Physics-based Data-informed Prediction of Vertical, Catenary, and Stepped Riser Vortex-induced Vibrations

Semi-empirical models serve as current state-of-the-art prediction technologies for vortex-induced vibrations (VIV). Accu -rate prediction of the flexible body's structural response relies heavily on the accuracy of the acquired hydrodynamic coeffi-cient database. The construction of systematic databases from rigid cylinder forced vibration experiments not only requires an extensive amount of time and resources but also is a virtually impossible task, given the wide multidimensional space the databases span. In this work, we improve the flexible cylinder VIV prediction by machine learning the hydrodynamic databases using measurements along the structure; such a methodology has been proven effective for vertical flexible risers in uniform and sheared flows using vibration amplitude and frequency data. This work demonstrates the effectiveness of the framework on flexible vertical risers in a stepped current and flexible catenary risers (with the catenary plane parallel or at an oblique angle with respect to the incoming flow). Moreover, the framework is applied to stepped (two-diameter) risers undergoing dual-frequency vibrations. Last, but not least, the framework is extended to using only sparse strain sens -ing. The predicted VIV responses using the learned hydrodynamic coefficient databases are compared with experimental observations.