Simulation of free-surface flow using a boundless grid

Physically based fluid simulation in recent years has been successful for small-scale fluids such as liquid in a cubic cavity. However, for boundless free-surface flow of large scale and irregular area, there is a critical trade-off between simulation efficiency and accuracy because of the restriction of the traditional regular computational grids. This paper introduces boundless computational grids based on hierarchical runlength encoding to simulate large-scale free-surface flow. We first modeled the free-surface flow with a lattice Boltzmann method, and calculated the surface curvature in the update process. We then introduced an effective method with which to calculate the surface curvature according to the surface tension so that the surface detail was enhanced, and obtained the curvature of each surface effectively as it was extracted. Furthermore, we dynamically compressed and indexed the computational cells with the improved hierarchical run-length encoding algorithm, so that the grid expands dynamically according to the fluid flow and the computational resources used were proportional to the volume of the fluid. Finally fluids in different situations were simulated realistically. The proposed method makes the best of the computational resources to perform the simulation with high resolution, and dynamically allocates resources so that the fluid can expand in random directions without boundaries, which is suited to the simulation of large-scale visual scenes.