Modeling subgrid-scale forces by spatial artificial neural networks in large eddy simulation of turbulence
Chenyue Xie, Jianchun Wang, E Weinan
Abstract
Spatial artificial neural network (SANN) models are developed for subgrid-scale (SGS) forces in the large eddy simulation of turbulence, whose input features are based on the derivatives of the filtered field at different spatial locations. The correlation coefficients of SGS forces predicted by the SANN models can be made larger than 0.99, much higher than that of the traditional gradient model. The SANN models perform better than the dynamic Smagorinsky model and the dynamic mixed model in spectrum prediction and in velocity and instantaneous flow structure statistics.
Topics & Concepts
TurbulenceLarge eddy simulationArtificial neural networkScale (ratio)Statistical physicsField (mathematics)Spatial correlationFlow (mathematics)Computer scienceMathematicsPhysicsMechanicsArtificial intelligenceStatisticsQuantum mechanicsPure mathematicsFluid Dynamics and Turbulent FlowsModel Reduction and Neural NetworksHeat Transfer Mechanisms