Presentations

Abstract

The tilt of floating wind turbines induces a production loss due to the misalignment with the flow, a potential production loss/gain due to the vertical motion of the nacelle, and a production gain due to the deflection of the wake. Standard wake modelling methods typically don't include the tilt effect when calculating production estimates. In some floater concepts, the turbine is eccentrically positioned relative to the floater's flotation center leading to a variation of the turbine's tilt with wind direction and wind speed. As the turbine tilts backward, the rotor area facing free wind is reduced and the wake is deflected slightly upwards, both effects impacting the production. Moreover, the hub height either increases or decreases depending on the turbine being upwind or downwind of the flotation centre. Consequently, the goal of this study is to assess the combined impact of a wind-speed and direction-dependent tilt of a wind turbine eccentrically mounted on a floating substructure for a mid-size floating wind farm. We compare the results to a case where the impact of the tilt is disregarded and to a centrally mounted tower. We also evaluate the variability in the production obtained from different wake deficit and deflection models and from a Reynolds-averaged Navier-Stokes (RANS) simulation. To assess the net impact of the tilt on power production, we first built an aero-elastic SIMA model of a generic 15 MW turbine centrically and centrally positioned relative to the center of flotation. By prescribing different inflow conditions, we construct multi-dimensional power and thrust curves relating the power and the thrust coefficients, including the effect of tilt, to wind speed and wind direction. A power and thrust curve without floater tilt (only rotor tilt) is used as a reference. The multi-dimensional power curves are applied in PyWake for the wake modelling calculation of a mid-size wind farm. The annual energy production with and without the effect of tilt is calculated using different wake models, namely the Bastankhah model, Fuga, and the Gaussian TurboPark, combined with the Jimenez vertical deflection model. All models are run with their respective default wake expansion parameter and wake superposition model. The Jimenez deflection model is also run with its default expansion parameter. In parallel, to better understand the behaviour of the Jimenez deflection model, we use the RANS solver PyWakeEllipSys to simulate a row of five 15 MW turbines. Each turbine is separated by 4.2 RD and tilts according to the tilt-dependent power and thrust curves. The tilt angles are prescribed from a PyWake Fuga simulation of the five-turbine case. Preliminary results show that in the five-turbine case, all models, except the Gaussian TuroPark, result in an overall reduction in power when comparing the tilted case to the no-tilt case. In the more realistic wind farm case, all models show an overall loss due to turbine tilt, but with a reduction of the loss due to wake deflection.