Abstract
A real time fluid simulation framework based on DirectCompute acceleration was proposed to solve problems of low simulation efficiency that caused by the increasing fluid particle size and long processing time that caused by the complex solid-liquid boundary. In the simulation, a grid local collision detection method was adopted to speed up the search for the solid-liquid boundary. Fluid calculation, collision detection and surface reconstruction were computed parallel in the GPU with DirectCompute technology to accelerate the SPH fluid simulation. It could optimize the storage structure of fluid simulation, and save the storage space. Experimental results show that the framework can effectively improve the fluid simulation speed with more complex static scene, and can show the details of the fluid spray.
| Original language | English |
|---|---|
| Pages (from-to) | 2467-2475 |
| Number of pages | 9 |
| Journal | Xitong Fangzhen Xuebao / Journal of System Simulation |
| Volume | 28 |
| Issue number | 10 |
| State | Published - 8 Oct 2016 |
| Externally published | Yes |
Keywords
- Collision detection
- Fluid simulation
- GPU acceleration
- Smoothed particle hydrodynamics