Hybrid Modeling Technique for Large Eddy Simulations

In this paper a multilayer feed-forward neural network, previously offline trained, is used in substitution of a Bardina’ scale similar model- (BFR) [1] Sub-Grid Scale (SGS) model in Large Eddy Simulation of a channel flow at Re τ = 180. Results show the ability of neural networks to identify and reproduce the highly non-linear behavior of turbulent flows.