WIND FARM WAKE MODELING AND LAYOUT OPTIMIZATION USING NUMERICAL TOOLS

The thesis addressed the issue of the wakes and their aerodynamical losses generated by wind farm turbines from two main perspectives: the potential microclimatic impacts they may cause in the surrounding environment (particularly if directed towards the mainland in the case of large offshore wind farms) and the degradation in performance and Annual Energy Production (AEP). The proposed methodology involved two software tools, PyWake and FLORIS, developed respectively by DTU and NREL, whose potential applications have been highlighted. These software tools implement a range of different analytical wake models and are capable of simulating wake propagation and assessing the resulting downwind velocity deficits comparing their behaviours. Once a case study has been selected, the potential impacts of the wake propagation have been evaluated: the microclimatic impacts aspect has been tackled by qualitatively and quantitatively comparing the velocity losses caused by wakes for each of the five wake models selected across both software tools. The results highlighted substantial differences between some of the presented models, confirming the need for benchmarking by other tools to validate the correct wake behaviour, but, overall, a limited impact was observed for every simulation. Regarding the optimization aspect, starting from the existing YawOptimizer function in FLORIS, a yaw control strategy algorithm has been implemented in PyWake, aimed at finding the optimal yaw angles in terms of energy production on an annual basis, to leverage this strategy with the various wake models within the PyWake library. Specific yaw angles can reduce the wake influence on the performance of downwind turbines, thereby enabling greater overall power extraction and, consequently, higher annual energy production. In this case as well, the improvements have been evaluated based on the selected wake model, yielding different results depending on the typical wake behaviour of each model: while some models have shown excellent results in terms of production increase, demonstrating how wake deficit can be impactful, others do not appear to be very sensitive to yaw control, which is why a benchmark to verify the proper behaviour of the wake is considered necessary.