Advances in semi-empirical time domain modelling of vortex-induced vibrations
MetadataShow full item record
- Institutt for marin teknikk 
Most of today’s prediction tools for vortex-induced vibrations (VIV) operate in the frequency domain, where analysis is restricted to linear systems and steady flows. These modelling limitations provide an unsatisfactory representation of realistic operational conditions faced by slender marine structures. For instance, marine risers may experience: waves and current loading combined, non-linear soil-pipe interaction, and top end excitation. In time domain, structural non-linearities and unsteady flows can be accounted for in simulation. However, good quality response estimates require that the hydrodynamic loads are properly modelled. With basis in the PhD work by Mats J. Thorsen, the present thesis aims to develop semi-empirical time domain load models with the capacity of handling a variety of complex phenomena associated with VIV of flexible pipes in particular. A goal has been to avoid unjust sacrifices with respect to the application range, computational efficiency, and model complexity, in the pursuit of accurate predictions. The overall objective is to provide good alternatives to existing methods for VIV assessment, suitable for engineering applications. Four fluid load models were developed in order to simulate: pure in-line VIV (I), stochastic cross-flow VIV (II), combined wave and current induced responses in crossflow and in-line directions (III), and higher harmonic responses (IV). Although these response phenomena are quite different, the application ranges associated with each of the load models are not necessarily mutually exclusive. For instance, proposal IV should be fully capable of predicting phenomenon III, as well as the amplitude modulations and the frequency variations that are associated with phenomenon II. The fluid load models where formulated as a sum of damping, added mass and vortex shedding terms, where a synchronization model was utilized to connect vortex shedding and cylinder motion. The load models were combined with linear finite element models of flexible pipes to simulate VIV. The accuracy of the response predictions were assessed by comparison with various experimental data, and the results were in general satisfactory. Riser VIV in uniform current, and in irregular waves and uniform current combined, was successfully modelled for the same choices of empirical parameters. By including one additional term to the fluid load model, both fundamental and higher harmonic VIV was realistically captured in simulation of flexible pipes in uniform current. Conclusions were made from an engineering perspective, where there is a preference for simple models containing few empirical parameters. In this context, the stochastic load model (II) did not improve accuracy to an extent that justifies for the complexity associated with the stochastic synchronization model. The other load models were considered to be of significant value, as they account for phenomena that have a major impact on fatigue damage, still remaining reasonably simple.