The wind farm as a sensor in highly complex terrain

Abstract This paper demonstrates the wind farm as a sensor methodology at an onshore site in highly complex terrain. The method provides a site-specific engineering flow model, by using the whole farm as a distributed sensor. Two years of operational turbine data, synthesised to long term observations, are available. For the site-adaption, an engineering wake model is augmented with parametric correction terms that model the heterogenous intra-farm flow field caused by the terrain at the site. This background field, together with the tuning parameters of the engineering wake model, are simultaneously learned through the observational data. High-fidelity CFD RANS simulations from different directions provide an independent source for the intra-farm field caused by the terrain. The data-driven flow field, as well as the CFD simulations, reveal the presence of significant orography-induced effects throughout the site.