Dynamic particle coupling for gpubased fluid simulation.pdf

Dynamic Particle Coupling for GPU-based Fluid Simulation Andreas Kolb, Nicolas Cuntz Computer Graphics Group, University of Siegen, Germany Email: {andreas.kolb,nicolas.cuntz}@uni-siegen.de Abstract The main research approaches in Computational Fluid Dynamics (CFD) are grid-based (Euler) or based on particle motion (Lagrange). For interactive fluid simulation, tech- niques have been developed to use the Graphics Processing Unit (GPU) to speed up the computation using an Eulerian approach. This paper describes an approach for setting up a Lagrangian particle based fluid simulation on the GPU. This builds upon earlier results on simulation of uncoupled particles. The major contribution of this work is a new approach for modeling dynamic particle coupling solely based on individual particle contributions. This technique does not need global sorting or an explicite solution of the n-nearest neighbor problem. 1 Introduction The simulation of fluid is important, e.g. in modeling physical systems (climate, ocean), but also in interactive computer graphics applications. Reeves [Ree83] used particle systems in the context of the motion picture Star Trek II. Further applications are the simulation and animation of soft objects [DG96] and the control of implicit surfaces [WH94]. In Computational Fluid Dynamics (CFD) one main research approach is grid-based (Euler). In contrast to the Eulerian approaches, the particle-based approach (Lagrange) makes mass conservation equations and convection terms dispensable. This reduces the complexity of the simulation and the particles can directly be used to render the fluids surface. Smoothed particle hydrodynamics (SPH) was introduced by Ginhold and Mon- aghan [GM77]. SPH models the dy