Wind turbine wake modelling including heterogeneous background flows
J.A. Kuleta (TU Delft - Aerospace Engineering)
S.J. Watson – Mentor (TU Delft - Aerospace Engineering)
Paul van der Laan – Mentor (Technical University of Denmark (DTU))
Mads Mølgaard Pedersen – Mentor (Technical University of Denmark (DTU))
Sam Williams – Mentor (RWE Offshore Wind GmbH)
D. Zappalá – Graduation committee member (TU Delft - Aerospace Engineering)
Alfredo Peña – Graduation committee member (Technical University of Denmark (DTU))
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
As offshore wind deployment continues to grow, accurate power prediction requires wake models that can represent realistic atmospheric conditions. Most steady-state engineering wake models assume uniform inflow across a wind farm, limiting their accuracy in coastal regions where wind speed and direction vary over relatively short distances. This thesis investigates the impact of spatial flow heterogeneity on wake model performance using a novel coupling between Whiffle-LES and PyWake, in which high-fidelity large-eddy simulations (Whiffle-LES) provide spatially varying inflow conditions. The approach is applied at two contrasting sites: the coastal Murakami Tainai planned wind farm in Japan and the operational Arkona offshore wind farm in the Baltic Sea.
The results show that the spatio-temporal resolution of the background field influences model performance. At Murakami Tainai, accounting for local variations in wind speed and direction significantly improves predictions of spatial power production patterns, while streamline-based wake propagation offers only limited benefit over conventional straight-line propagation. At Arkona, spatial heterogeneity has little impact on annual energy production and wake-efficiency estimates, although field-dependent inflow descriptions can capture spatial production patterns relevant for power forecasting and wind farm control. At both sites, averaging multiple inflow realizations based on local turbine conditions reproduces wind farm power trends nearly as well as field-dependent approaches.
The influence of spatial variability is found to be secondary to the wake formulation itself. Under stable atmospheric conditions, engineering wake models consistently overpredict wind farm power, underestimating wake deficit lengths. Their performance improves considerably under unstable conditions. Among the wake models considered, the TurboGaussian model provides the best agreement with both SCADA measurements and high-fidelity Whiffle-LES simulations. Overall, the results indicate that further improvements in engineering wake modelling require not only more realistic inflow descriptions but also wake formulations that better account for atmospheric conditions.