Implementation and Optimization of Lattice Boltzmann Method for Fluid Flow on GPU with CUDA

Abstract

Abstract In recent years, the computational fluid dynamics has been widely used in the aviation, weather, chemical industry and so on. The demand on computing power for modern computational fluid dynamics is getting higher and higher. Aiming at the features of the LBM (lattice Boltzmann method), we put forward a parallel data structure layout of the D2Q9 model and related algorithm based on CUDA(compute unified device architecture).Taking the plane Poiseuille flow as a test example, we implement the parallel computation of the LBM based on LBGK (lattice Bhatnagar-Gross-Krook) model on GPU(graphic processing unit).By comparing the results obtained on a NVIDIA GeForce GTS 450 GPU with those on a Intel Core i5-760 quad-core CPU, both have good agreement and the highest speedup of GPU is about 102 times faster than that of CPU. The result indicates parallel computation of the LBM on GPU is completely feasible, which provides a very effective way to solve the complex problem of the modern computational fluid dynamics on a low cost computer.