How accurately do engineering methods capture floating wind turbine performance and wake? A multi-fidelity perspective

Abstract. Despite an increasing number of experimental and numerical studies, the influence of platform motion on wake dynamics (wake recovery and turbulence production) in floating offshore wind turbines is still an open research question. In particular, efforts are being made to understand the accuracy of numerical models in use so far for fixed-bottom turbines when they are applied to floating configurations. Similarly to what has been done in IEA's OC6 task, in this work a multi-fidelity approach is leveraged to investigate the capabilities of engineering models to capture the wake dynamics of a wind turbine model under imposed motion. Differently from previous studies, however, many more different operating conditions have investigated, including surge, pitch, yaw and wind-wave misalignment cases; moreover, numerical methos are here consistently applied to the same test cases, which are part of the first experimental round of the NETTUNO project. More specifically, Free Vortex Wake (FVW), Actuator Line Model (ALM) and blade resolved CFD simulations have been benchmarked and their capabilities in predicting the mean wake response and the onset of velocity oscillations in the wake of a floating wind turbine were evaluated. Results showed that, up to 5D and in the operating conditions tested, platform motion has limited impact on the wake in terms of wake deficit. However, significant velocity oscillations are observed at a platform reduced frequency of 0.6 which could be detrimental for downstream machines. An investigation of the vortex structures in the wake showed that these velocity oscillations might be caused by the interaction of vortex structures generated under sinusoidal platform motion rather than by unsteady aerodynamic response of the rotor. FVW methods, if properly tuned, can correctly capture the wake response up to 3D from the rotor, but to simulate the wake response up to 5D, higher-fidelity methods are required. Significant improvements are achieved with ALM CFD simulations, even though an URANS approach might struggle to correctly predict the wake dissipation due to the interaction between the free-stream turbulence and wake.