26260

gLBM: A GPU enabled Lattice Boltzmann Method Library

Aaron Bray, Rachel B. Clipp, M. Umar Qureshi, Sorin Mitran, Andinet Enquobahrie
Kitware, Inc., Carrboro, NC 27510
Journal of Open Source Software, 7(70), 2555, 2022

@article{bray2022glbm,

   title={gLBM: A GPU enabled Lattice Boltzmann Method Library},

   author={Bray, Aaron and Clipp, Rachel B and Qureshi, M Umar and Mitran, Sorin and Enquobahrie, Andinet},

   journal={Journal of Open Source Software},

   volume={7},

   number={70},

   pages={2555},

   year={2022}

}

Lattice Boltzmann Methods (LBM) are a class of computational fluid dynamics (CFD) algorithms for simulation. Unlike traditional formulations that simulate fluid dynamics on a macroscopic level with a mesh, the LBM characterizes the problem on a mesoscopic level applied to a grid discretization. LBM solves the fluid density problem with collide and stream (relaxation) processes. This approach has several advantages, including its adaptability to numerous fluid domains (i.e., vapours, liquid droplets), complex boundaries, irregular interior geometries, and parallelization. Traditional CFD methods are limited in the ability to parallelize the algorithm; however, the LBM algorithm discretization can be easily parallelized both for CPUs and GPUs. This enables fast fluid solutions for complex fluid domains. There are limitations associated with the LBM, including high Mach number applications. However, active research is addressing these limitations.
No votes yet.
Please wait...

* * *

* * *

* * *

HGPU group © 2010-2022 hgpu.org

All rights belong to the respective authors

Contact us: