Wall-modeled LES (WMLES) is promising to simulate a turbulent flow at high Reynolds number with a reasonable cost because the required computational cost of large-eddy simulation in the near-wall region is proportional to approximately square of the f...
Wall-modeled LES (WMLES) is promising to simulate a turbulent flow at high Reynolds number with a reasonable cost because the required computational cost of large-eddy simulation in the near-wall region is proportional to approximately square of the friction Reynolds number (Reτ). An equilibrium stress model is the most widely used method due to the high efficiency. However, this method is still required to enhance the accuracy and applicability because of the limitations of the equilibrium assumption. In the present study, an artificial neural network (ANN) is used to obtain the wall shear stress for WMLES. The proposed method shows good prediction on the profiles of mean velocity and Reynolds stresses compared to equilibrium model when direct numerical simulation dataset of turbulent channel flows is trained at Reτ=395, 540, 930 and 2000. In addition, the present method predicts well the turbulent statistics at Re<SUB>τ</SUB>=640 and 5200, which are untrained by the ANN.