Comparison of CFD Predictions-of Multi-Mode Vortex-Induced Vibrations of a Tension Riser With Laboratory Measurements

Abstract

This paper compares previously unpublished laboratory measurements of the vortex-induced vibrations of a vertical tension riser with predictions of five CFD-based riser codes. The experiments were carried out with a 13m long, 28mm diameter model riser of which the upper 7m was in air, and the remainder in water with a uniform flow generated by carriage motion at speeds up to 1m/s. In-line and cross-flow responses were computed from measurements of the curvature of the riser at 32 stations over its length. Key results are compared with blind CFD predictions in which the flow is computed in two dimensions on each of a number of horizontal planes. The CFD codes used in this exercise represent a wide range of approaches, represented by (1) a vorticity-stream function method using a finite volume technique for integrating the vorticity transport equation, (2) finite elements with linear interpolation functions on a triangular grid, (3) a pressure-correction method in a finite-volume integration scheme on hybrid unstructured grids, (4) a discrete vortex formulation incorporating the growing core size or core spread method to model diffusion of vorticity, (5) a velocity-vorticity method using a hybrid Eulerian–Lagrangian vortex-in-cell method with LES, diffusion computed on a grid, and convection of vorticity by discrete vortices. In individual test cases, maximum computed cross-flow responses were between 23% and 160% of measured values, and there was a much wider range in ratios of predicted to measured curvatures. Overall, average ratios of predicted to measured cross-flow responses were between 59% and 106% for the 5 different codes, and those for maximum in-line deflections between 71% and 109%. Predicted curvatures in either direction were on average between 30% and 170% of the measured values.