A comparative study of bounce-back and immersed boundary method in LBM for turbulent flow simulation

In this study, we compared the modified bounce-back (BB) method and immersed boundary (IB) method for the treatments of no-slip boundary condition based on the Lattice Boltzmann Method (LBM) for turbulent flow. The benchmark case of flow past a square cylinder confined in a rectangular duct has been simulated with two schemes at Re d = 3000. The turbulent flow was modeled using Large Eddy Simulation (LES). The conventional Smagorinsky subgrid scheme was used to resolve the small-scale motions. The proposed algorithm is parallelized to run on NVIDIA Tesla P100 Graphics Processing Unit (GPU) using Compute Unified Device Architecture (CUDA) programming language. The results obtained from the bounce-back and immersed boundary method has been validated with the experimental and LES data reported in the literature for different turbulent statistics including time-averaged velocities, root-mean-square velocity fluctuations, and Reynolds shear stress. Later, the comparison between the two approaches discussed in terms of accurate predictions for the drag coefficient. The computational efficiency on GPU for both methods has also been reported. [copyright information to be updated in production process]