Model-form uncertainty quantification in RANS simulations of wakes and power losses in wind farms.

Reynolds-averaged Navier-Stokes (RANS) is one of the most cost-efficient approaches to simulate wind-farm-atmosphere interactions. However, the applicability of RANS-based methods is always limited by the accuracy of turbulence closure models, which can introduce various uncertainties into the models. This study aims to address this issue in RANS simulations of wakes and power losses in wind farms. For this purpose, the performance of different RANS models is first tested with a large-eddy simulation (LES) of an idealized wind farm. It is found that, for the particular case considered here, the realizable k-\epsilon model can yield a relatively more accurate prediction for the mean velocity, turbulence intensity, and power losses within the wind farm. However, it fails to accurately predict the structure of turbulence in the wake region. Inspired by the obtained results, the model-form uncertainty associate with the turbulence model is investigated and quantified by perturbing the Reynolds stress tensor. The focus is placed on perturbing the shape of the tensor represented by its eigenvalues. The results show that the perturbed RANS model successfully estimates the region bounding the LES results for quantities of interest. Finally, the selection of introduced-perturbation amount and its effects on the model predictions are discussed.