Two grid refinement methods in the lattice Boltzmann framework for reaction-diffusion processes in complex systems

This paper studies the optimisation of a numerical model and a computer code to solve numerically reaction-diffusion processes in environmental or biological systems with complicated geometries and mixtures of reactions including time and spatial scales extending over several order of magnitude. In particular, we consider different grid refinement techniques in the framework of a lattice Boltzmann solver for reaction-diffusion systems. Two new grid refinement methods are proposed, which are both quantitatively good. The first method is based on the matching of the concentration profiles and fluxes across two adjacent sub-domains, while the second method is based on nested subgrids. The focus of our study is the trade off between accuracy and CPU time. We show how the different parameters of the method, such as the refinement factors, the location of the boundary between different grids or coupling methods at the interface affect the quality of the numerical solution and the efficiency of the method.