Using Parallel Computing Language to Enhance our Understanding of Lattice Boltzmann Model

- 104632
Poster
Favorite this paper
How to cite this paper?
Abstract

The Lattice Boltzmann method (LBM) has enjoyed its popularity as a promising tool for simulating complex fluid flows. The accuracy and the stability of LBM has been improved by various techniques throughout its development. In this study, we use LBM to simulate the interaction between air flow and the flapping of wings of hawk moths. The high frequency of the flapping movement requires the algorithm to be robust for high Reynolds numbers. This poster shows preliminary results comparing the stability and the accuracy of the single-relaxation-time, the two-relaxation-time (TRT) and the multiple-relaxation-time (MRT) schemes. We implemented these algorithms in high-performance parallel GPU languages CUDA and WebGL. The code is designed to be highly adaptive and user-friendly. These simulations will provide insights and guidance to our collaborators who design experiments in wind tunnels to study the movement of insects’ wings.

Institutions
  • 1 Georgia Institutte of Technology
  • 2 Georgia Technological Institute
Track
  • Scientific Computing in Computer Science
Keywords
Lattice Boltzmann method (LBM)
Computational Fluid Dynamics (CFD)
GPU-CUDA
WebGL