A robust lattice Boltzmann method for parallel simulations of multicomponent flows in complex geometries. (10th August 2017)

• Record Type: Journal Article
• Authors: Zudrop, Jens; Masilamani, Kannan; Roller, Sabine; Asinari, Pietro
• Is Part Of: Computers & fluids. Volume 153(2017)
• Page Start: 20
• Page End: 33
• Publication Date: 2017-08-10
• Subjects: Lattice Boltzmann -- Multicomponent flow -- Maxwell–Stefan diffusion -- High performance computing -- Massively parallel simulations -- Octree; Fluid dynamics -- Data processing -- Periodicals; 532.050285
• DOI: 10.1016/j.compfluid.2017.04.021
• Languages: English
• ISSNs: 0045-7930

Abstract

Highlights: We present a robust and efficient Lattice Boltzmann Method for non-ideal mixtures on octree meshes. We analyze boundary conditions for the presented model. Well-posedness of the implicit-to-explicit variable transformation of the Lattice Boltzmann Method is shown. Massively parallel simulation of a realistic mixing process in a porous media with billions of species elements. We demonstrate high computational performance on a high ranked TOP500 system. Abstract: In this work, we are concerned with robust parallel simulations of multicomponent flows in complex geometries by an extended lattice Boltzmann method. In Zudrop et al. (2014) [22] we presented a model for the incompressible Navier–Stokes and Maxwell–Stefan diffusion equations. Here, we prove that the implicit-to-explicit variable transformation of the model is well-posed under weak constraints on the dimensionless Maxwell–Stefan diffusivities. Furthermore, we analyze various boundary conditions for the multicomponent lattice Boltzmann model. These results make the model robust and well suited for complex geometries, which allows us to study realistic, large-scale mass transport applications.