Predictability Study of Viscoelastic Turbulent Channel Flow Using Deep Learning

We tested Artificial Neural Networks (ANNs) to predict a fully-developed turbulent channel flow of a viscoelastic fluid in preparation for elucidating flow phenomenon and solving the difficulty in DNS (Direct Numerical Simulation) due to numerical instability of the viscoelastic fluid. Two kinds of ANNs (multi-layer perceptron (MLP) and U-Net) were trained using DNS data to predict conformation stress from given instantaneous field. The MLP showed accurate predictions and predictions got better with z-score normalization. ANN predicted accurately in near-wall region having coherent structures. In addition, we demonstrated that ANN get the nonlinear relationship between velocity gradient and viscoelastic stress partially.