Deep unsupervised learning of turbulence for inflow generation at various Reynolds numbers

•We applied an unsupervised deep learning algorithm to inflow generation of turbulent channel flows.•We combined GAN and RNN to generate 2D time-varying turbulent flows.•The trained network could generate the flows at various Reynolds numbers, outside of trained one.•We could achieve high statistical accuracy compared to DNS. A realistic inflow boundary condition is essential for successful simulation of the developing turbulent boundary layer or channel flows. In the present work, we applied generative adversarial networks (GANs), a representative of unsupervised learning, to generate an inlet boundary condition of turbulent channel flow. Upon learning the two-dimensional spatial structure of turbulence using data obtained from direct numerical simulation (DNS) of turbulent channel flow, the GAN could generate instantaneous flow fields that are statistically similar to those of DNS. After learning data at only three Reynolds numbers, the GAN could produce fields at various Reynolds numbers within a certain range without additional simulation. Eventually, through a combination of the GAN and a recurrent neural network (RNN), we developed a novel model (RNN-GAN) that could generate time-varying fully developed flow for a long time. The spatiotemporal correlations of the generated flow are in good agreement with those of the DNS. This proves the usefulness of unsupervised learning in the generation of synthetic turbulence fields.