Adaptive large-eddy simulations with a dynamic subgrid model based on local velocity fluctuation intensities

Abstract In this work, an improved adaptive approach for large-eddy simulations (LES) is introduced based on a dynamic adaptive mesh refinement (AMR) algorithm and a corresponding adaptive subgrid model embedded in such AMR process. The objective of the present work is to develop a turbulent-flow simulation method that dynamically allocates meshes of varying resolutions, which thereby significantly reduces the required computational resources of traditional LES methods. To achieve this, the local velocity fluctuation intensity is proposed as a criterion to determine whether the grid resolution needs to be changed. With this, the grid distribution is adjusted according to the motions of the local flow structures, enabling a more precise and cost-effective way to capture the multiscale evolution in turbulent flow fields. Meanwhile, with the updated position and velocity information, an adaptive subgrid model is constructed within the AMR framework to determine the behaviors of the unresolved structures smaller than the finest resolution of the AMR mesh. As a result, the developed dynamic remeshing and the corresponding adaptive subgrid model enhance the performance and efficiency of LES models. Through a number of validation examples, the accuracy, capability, and efficiency of the proposed adaptive LES methodology have been demonstrated.