3-D shear-layer model for the simulation of multiple wind turbine wakes: description and first assessment

Abstract

Abstract. The number of turbines installed in offshore wind farms has strongly increased in the last years and at the same time the need for more precise estimations of the wind farm efficiency too. In this sense, the interaction between wakes has become a relevant aspect for the definition of a wind farm layout, for the assessment of its annual energy yield and for the evaluation of wind turbine fatigue loads. For this reason, accurate models for multiple overlapping wakes are a main concern of the wind energy community. Existing engineering models can only simulate single wakes, which are superimposed when they are interacting in a wind farm. This method is a practical solution, but it is not fully supported by a physical background. The limitation to single wakes is given by the assumption that the wake is axisymmetric. As an alternative, we propose a new shear-layer model that is based on the existing engineering wake models but is extended to also simulate non-axisymmetric wakes. In this paper, we present the theoretical background of the model and four application cases. We evaluate the new model for the simulation of single and multiple wakes using large-eddy simulations as reference. In particular, we report the improvements of the new model predictions in comparison to a sum-of-squares superposition approach for the simulation of three interacting wakes. The lower deviation from the reference considering single and multiple wakes encourages the further development of the model and promises a successful application for the simulation of wind farm flows.

A simulation study demonstrating the importance of large-scale trailing vortices in wake steering

“…In the upper FLORIS figure, you can see that FLORIS assumes that both yawing and translating the front turbines will cause the power deficit far downstream to be both shifted a little bit, and in a similar way. However, SOWFA shows that while the FLORIS prediction of the combined partial wake is not as accurate as its prediction of the incoming single wake (as could be expected from Trabucchi et al, 2017), the substantial change in the yawed case is much more different than FLORIS' prediction. Note for example the minimum of the "profile" of the yawed wake far downstream is further deflected than the incoming steered wake.…”

“…FLORIS' wake combination model, as described earlier, is based on sum of squares. Given that Trabucchi et al (2017) show that sum-of-squares superposition, as is used in FLORIS, has an error in prediction relative to newer methods, this could be the explanation for the discrepancy; i.e., partial wake overlap leads to an apparent secondary steering and improved wake combination models in FLORIS would resolve the discrepancy.…”

“…Overlapping wakes are combined using a sum-of-squares approach that has been used previously in literature (Katić et al, 1986). It is important to note that there are alternative methods to combine wakes as wake superposition is an ongoing research topic (Machefaux and Mann, 2015;Trabucchi et al, 2017).…”

Theory and Verification of a new 3D RANS Wake Model

“…One branch was continued by TNO/ECN (The Energy Research Center of the Netherlands), and it resulted in the WakeFarm presented by Schepers (2003), and FarmFlow model presented in Eecen et al (2011). Renewed interest in mid-fidelity models has recently led to the independent development of several new models 40 in this group, like those presented by Trabucchi et al (2017) and Martinez-Tossas (2019).…”