Physical and Computational Scaling Issues in Lattice Boltzmann Simulations of Binary Fluid Mixtures

TL;DR

This paper discusses physical and computational scaling challenges in lattice Boltzmann simulations of binary fluid mixtures, emphasizing the importance of physics resolution and numerical accuracy in large-scale and anisotropic flow simulations.

Contribution

It identifies key physical and computational scaling issues in lattice Boltzmann methods and proposes strategies for effective simulation parameter selection and error minimization.

Findings

Physical scaling affects how simulation parameters relate to real-world fluids.

Anisotropic systems require adapted simulation geometries.

Strategies to minimize numerical errors in large-scale shear flow simulations.

Abstract

We describe some scaling issues that arise when using lattice Boltzmann methods to simulate binary fluid mixtures -- both in the presence and in the absence of colloidal particles. Two types of scaling problem arise: physical and computational. Physical scaling concerns how to relate simulation parameters to those of the real world. To do this effectively requires careful physics, because (in common with other methods) lattice Boltzmann cannot fully resolve the hierarchy of length, energy and time scales that arise in typical flows of complex fluids. Care is needed in deciding what physics to resolve and what to leave unresolved, particularly when colloidal particles are present in one or both of two fluid phases. This influences steering of simulation parameters such as fluid viscosity and interfacial tension. When the physics is anisotropic (for example, in systems under shear) careful adaptation of the geometry of the simulation box may be needed; an example of this, relating to our study of the effect of colloidal particles on the Rayleigh-Plateau instability of a fluid cylinder, is described. The second and closely related set of scaling issues are computational in nature: how do you scale up simulations to very large lattice sizes? The problem is acute for systems undergoing shear flow. Here one requires a set of blockwise co-moving frames to the fluid, each connected to the next by a Lees-Edwards like boundary condition. These matching planes lead to small numerical errors whose cumulative effects can become severe; strategies for minimising such effects are discussed.