Fig. 1: Sketch of Physics-Informed Neural Network for LBM-based wall model. | Communications Physics

Fig. 1: Sketch of Physics-Informed Neural Network for LBM-based wall model.

From: Physics informed data-driven near-wall modelling for lattice Boltzmann simulation of high Reynolds number turbulent flows

Fig. 1

Diagram of the physics-informed neural networks for wall modelling and implementation for the lattice Boltzmann method trained by IDDES data. a the training data is obtained from the upper and lower near-wall region of a periodic channel flow simulation. We use NN to predict the shear velocity which can be used by the wall model. b LES based LBM channel flow simulation is performed; on the first layer of the wall, we use the LBM data as the input. With the help of the pre-trained PINNs wall model, we can predict LBM shear velocity in order to calculate the resistance wall force Fw for the wall model.

Back to article page