This paper presents a new quantitative approach of defining current profiles for application to riser fatigue assessment using profile clustering methods.
The analysis presented here was undertaken using a one year long deepwater current profile dataset from the Gulf of Mexico. The data included near full water column measurements in 3250m water depth at one hour intervals, providing nearly 9000 individual profiles. Riser fatigue damage for each profile had been previously computed as part of the Worldwide Approximation of Current Profiles (WACUP) joint industry project.
The new assessment described in this paper applies clustering methods not considered in WACUP, including the K-Means Algorithm (KMA) and Maximum Dissimilarity Algorithm (MDA). These both demonstrate superior performance compared to a much simpler direct method of characterisation. Features of the KMA and MDA methods are contrasted, within the context of previously published application to ocean wave data.