Dynamic mode decomposition analysis for Savonius wind turbine

Abstract

Wind turbines have been used to harvest clean energy for many years. However, wind turbine simulation is extremely expensive as it requires a high computational cost. Consequently, many studies have proposed different methods to reduce the computational time of wind turbine simulation. Reduced Order Methods (ROMs) show their capability to predict the flow field in many cases but have not been applied to a Savonius wind turbine. This study is intended to utilize the Dynamic Mode Decomposition (DMD), one of the ROM models, for the first time to predict the wake of a Savonius wind turbine. In this regard, two types of predictions are conducted. First, the main variables of the flow field are calculated for interpolating the results from the numerical simulation. The results show a 52 percent reduction in the run time with a mean R2 equal to 0.95. Meanwhile, the time spent in the DMD method is negligible. Second, the first four cycles of a wind turbine are used for the prediction of the next four cycles. The findings for this case are also very accurate, and the DMD shows its ability to predict the wake of a Savonius wind turbine.