Incorporating particle motion into an ADF for fast coupling of fluids with rigid and deformable solids

Abstract

We present a new method for the fast simulation of interactions between fluids and solids by incorporating particle-based water flow into an adaptive signed distance field. In some previous methods, the motion of every water particle is checked when simulating the collision with the solid in the coupling process, and the computational cost becomes very great as the number of particles increases. If only the particles on the leaf nodes surrounding the solid are considered, this reduces the computational cost of collision detection, but some collisions may not be detected. This may lead to the "tunneling" artifact, in which particles with high velocities skip across the layer of leaf nodes. This paper addresses the problem by (i) considering particles only on the leaf nodes in the adaptive structure to improve the processing time required for the water-solid coupling and (ii) considering the water flow to avoid the tunneling artifact by incorporating particle motion into the tree structure of the adaptive signed distance field. Our method can be computed in parallel, and experimental results show that it outperforms previous methods while producing animations that are largely free of artifacts. Copyright