アンサンブルカルマンフィルタを用いた乱流モデルのパラメータ最適化

抄録

An optimization approach is presented that can be used to determine optimal parameter sets of a turbulence model in order to evaluate the Reynolds stress correctly. The approach employs the Ensemble Kalman filter (Enkf), which is one of data assimilation techniques and build by remodeling Kalman filter to estimate the initial state of nonlinear systems based on the observation state such as experimental result or high-accuracy simulation results. In this study, a target of the optimization was the parameters of k-ω 2 equations model which is one of the most famous turbulence models and widely used to design fluid machineries and analyze flow fields. Enkf was combined with URANS performed in a 2-dimensional flow field, and the observation state was the time averaged LES result which performed in 3-dimensional flow field and averaged into 2-dimensional flow field. Above flamework was applied to a transonic flow though the T106 A turbine cascade. The evaluation of this optimization approach was carried out with 3-dimentional URANS, and it revealed that the flow on the URANS using the optimized parameters showed a good agreement with the unsteady turbulent flow such as eddies that were not able to be recognized in the flow filed of the URANS using conventional parameters, so that the effect of the optimization was proved.