Wind farm layout optimisation consists of a long chain of models propagating information to the final energy yield prediction. To reduce the number of input variables and reduce computational costs, it is necessary to finish the wind resource assessment (WRA) before the layout optimisation is started. The WRA uncertainty is therefore not propagated to the layout optimisation. Since flow modelling and wake effects constitute the two most significant uncertainties in energy yield predictions, the separation is hazardous. In a new paper Gomes da Silva, Regina de Andrade, and Zaparoli (2021) illustrates how the use of four different flow models lead to four very different layouts for a complex site in Brazil. Since the flow modelling is uncertain, which of the four modelled wind fields best represent reality is unknown. Various combinations of wind fields and layouts lead to more than a 13% difference in the predicted AEP. Gomes da Silva, Regina de Andrade, and Zaparoli (2021) thereby emphasise the danger of relying on only a single flow model for WRA.

Gomes da Silva, Adaiana F., Claudia Regina de Andrade, and Edson Luiz Zaparoli. 2021. “Wind Power Generation Prediction in a Complex Site by Comparing Different Numerical Tools.” Journal of Wind Engineering and Industrial Aerodynamics 216 (September): 104728. https://doi.org/10.1016/j.jweia.2021.104728.