A simple and efficient curved boundary scheme of the lattice Boltzmann method for Robin boundary conditions of convection–diffusion equations

Xinyuan Xie, Weifeng Zhao (Lead / Corresponding author), Ping Lin (Lead / Corresponding author)

Research output: Contribution to journalArticlepeer-review

Abstract

This paper is concerned with boundary schemes of the lattice Boltzmann method for Robin boundary conditions of convection–diffusion equations on curved boundaries. For such boundary conditions, all the existing boundary schemes suffer from the possibility that the denominator in the scheme may become zero, which will lead to numerical instability. To avoid this possibility, we propose a boundary scheme by approximating the gradient along the outgoing discrete velocity at the boundary with the given Robin boundary condition and the gradient at the interior point next to the boundary. With this approximated gradient at the boundary, the classical bounce back scheme for Neumann-type boundary conditions is employed to obtain the unknown distribution function at the interior point. The scheme obtained has the first-order accuracy for curved boundaries and its advantages are: (1) the scheme is simple in form so that it can be easily implemented; (2) it avoids the denominator in the scheme to be zero, and (3) the scheme is single-node, i.e., it only involves the information at the present point. Numerical examples demonstrate the designed accuracy and good stability of our scheme for complex boundaries.

Original languageEnglish
Article number107536
Number of pages7
JournalApplied Mathematics Letters
Volume122
Early online date22 Jul 2021
DOIs
Publication statusE-pub ahead of print - 22 Jul 2021

Keywords

  • Boundary scheme
  • Convection–diffusion equations
  • Lattice Boltzmann method
  • Robin boundary condition

Fingerprint

Dive into the research topics of 'A simple and efficient curved boundary scheme of the lattice Boltzmann method for Robin boundary conditions of convection–diffusion equations'. Together they form a unique fingerprint.

Cite this