Simulation and Visualization of Fire with the Lattice-Boltzmann Method

The realistic depiction of fire is a challenge for the disciplines of computational simulation and visualization alike. As the phenomenon is not fully understood yet, finding a suitable model for the lifelike simulation of fire is difficult. Therefore, most approaches concentrate on the creation of realistic imagery rather than strict physical accuracy. This thesis attempts a physically motivated simulation and visualization of freely fed fire. The implementation is based on a model presented in the paper “Simulating Fire With Texture Splats” of Wei et al. Fire and smoke are represented by a cloud of particles that are advected by a velocity field computed with the Lattice-Boltzmann method (LBM). Combined with the Smagorinsky subgrid correction, the fluid simulation is stable even for turbulent flows. Additionally, the LBM provides an easy way to integrate second-order boundary conditions and obstacles in the flow. In this way, the interaction of the airflow with objects and environmental influences like wind can be modeled. A temperature field is added to control the creation of smoke from the fire and buoyancy-induced particle movements. Supplemented with an OpenGL 3D engine, the results can be immediately visualized in real-time with interactive frame rates, using billboards as particle representations.