Simulation Of Gas-Liquid Mixture Flow Based On Lattice Boltzmann Method

Computer animation techniques have witnessed a surging development in the last decade,seeking realism by means of physically based approaches. Realistic simulations of naturalphenomena such as fluid motion, are made possible, in contrary to the conventional syntheticmethodology like noise generation. It was not until a decade and a half ago that the computergraphics community turned to a physically consistent way for fluid animation by solving theNavier-Stokes equations.The NS equations are a set of partial differential equations that govern fluid’s mass,momentum and energy from a macroscopic perspective, which are often difficult to solve dueto the coupling of the fields. For graphics purposes, incompressibility condition is usuallyassumed. Lattice Boltzmann method is a promising mesoscopic approach for fluid simulation.It is evolved from the lattice gas automaton and the kinetic theory of gases. By describing theevolution of a group of particle distribution functions, it can recover the weakly compressibleN-S equations and macroscopic fluid motion. In LBM, the distribution functions are storedon a lattice, hence the name, with discrete propagation directions. By propagating andapplying the collision operator on the distribution functions consecutively, fluid motion canbe simulated.A class of open problems in fluid simulation is multiphase flow with air-water mixture.The two phases with diverging properties, such as density contrast, are accompanied withcomplex interface motion, challenging traditional approaches, firstly by the need to trackinterface motion reliably, secondly by requiring proper interface boundary conditions. Aconventional free surface simplification deals with the difficulties by ignoring the gas phasecompletely and establishing pressure boundary conditions at the interface, while leavingbubble dynamics intractable. In LBM, intermolecular forces are introduced to account forair-water interaction so the solver can be unified without the need for interface tracking, butallows convenient simulation of complex multiphase phenomena including phase separation,bubble dynamics, and surface tension flow. The two phases with density contrast alsoapproximate free surface flow well.In this work, we aim to base a unified solver for air-water mixture on two-phase LBMmethods. We propose improvements for TP-LBM to enhance its native stability, and adopt aphysically based subgrid scale turbulence model to improve stability further in high Reynoldsnumber configurations when simulating interface turbulence. By our experiments, we showthat the framework solves two-phase flow with density contrast resembling air-water mixturein multiple settings, serving the purpose of making CG animations.