76th Annual Meeting of the Division of Fluid Dynamics
Sunday–Tuesday, November 19–21, 2023;
Washington, DC
Session J24: Energy: Wind Power Wakes, Control and Fluctuations II
4:35 PM–6:32 PM,
Sunday, November 19, 2023
Room: 150A
Chair: Ethan Lust, US Naval Academy
Abstract: J24.00004 : Effects of wind turbine thrust coefficient, turbine density, and incoming flow on wind farm wakes: A wind tunnel study.
5:14 PM–5:27 PM
Abstract
Presenter:
Wasi U Ahmed
(University of Texas at Dallas)
Authors:
Wasi U Ahmed
(University of Texas at Dallas)
Giacomo Valerio Iungo
(University of Texas at Dallas)
With the exponential growth of wind energy, wind farms can be built with a relatively small separation leading to mutual interferences, which are denoted as farm-to-farm interactions. Wakes generated by an upstream wind farm can affect power capture and fatigue loads of a downstream neighboring wind farm. Therefore, to predict wind farm interactions, it is instrumental to characterize and predict wind farm wakes for different thrust coefficients Ct and separation distance of the wind turbines under different incoming wind conditions. To this aim, an experiment was performed at the UT Dallas Boundary LAyer and Subsonic wind Tunnel (BLAST) using hot-wire anemometry (HWA). Porous disks with hub-height H = 8 cm, diameter D = 8 cm, and Ct = 0.55 or 0.85 were installed to reproduce wakes generated by different wind farms. The effect of streamwise spacing, Sx, and spanwise spacing, Sy, among adjacent disks, Ct, and incoming boundary layer is examined. The preliminary results show that despite having different Sx and Sy, for similar incoming wind conditions and Ct, the far wake (> 30D) velocity field in terms of mean streamwise velocity and turbulence intensity, TI, becomes self-similar. Further, vertical profiles at different downstream locations show that the variability of the mean wind speed and TI, just above the wind farm, depends strongly on Sx and Ct, while Sy has negligible effects. In addition, the wind farm wake velocity estimated by different engineering wake models is compared with HWA measurements. The results show that the Jensen model underestimates the wind-farm wake velocity deficit, while the turbulence optimized model with top-hat profile (TurbOPark) model predicts the wind farm wake closely with the HWA measurements. However, the Gaussian profile (TurbOGauss) model closely estimates the velocity deficit at the center of the disk columns but underestimates the wake velocity deficit at the radial locations between consecutive disks. Moreover, Frandsen's added wake turbulence model estimates the TI downstream of a wind farm with a very close approximation with the measurements when the wake expansion coefficient is set as a function of the local TI. Lastly, the wake recovery model by Emeis closely predicts the velocity recovery of wind farms with different configurations at downstream locations greater than 5D.