Data-Enhanced RANS Modeling of Unsteady Transitional Flows

This paper presents a training approach to improve the ability of a baseline RANS turbulence model to predict separation induced transitional flows over low Reynolds number airfoils and improve its prediction of the dynamic stall onset for pitching airfoils. We employ the learning and inference assisted by feature space engineering framework to train the underlying model parameters to predict the intermittency field that augments the baseline turbulence model. The training is performed using steady state airfoil cases at different angles of attack. Then, the trained model is applied to new steady state unseen cases with different airfoil sections and Reynolds number. Eventually, the trained model is used to simulate the flow over pitching airfoils. The results show improved ability of the trained model to capture flow transition and the laminar separation bubble for all training and testing static airfoil cases. However, the pitching airfoils simulations indicate the requirement for further improvements, particularly, enhancing the turbulent kinetic energy production within the laminar separation bubble.