GPU加速的二维流体实时流动仿真  

Real-time 2-D fluid simulation with a GPU acceleration

作　　者：佟志忠[1] 姜洪洲[1] 韩俊伟[1] 

机构地区：[1]哈尔滨工业大学机电工程学院,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2008年第3期278-284,共7页Journal of Harbin Engineering University

基　　金：教育部"新世纪优秀人才支持计划"基金资助项目(NCET-04-0325)

摘　　要：Stable Fluids是计算机图形学领域标准的流体仿真技术,但其存在一定的数值耗散.针对这一问题,将计算流体动力学领域(CFD)成熟的SMAC法移植到计算机图形学领域,结合CFD领域合理的边界条件处理及其相关技术进行二维流体的实时流动模拟.基于2种技术在可编程图形硬件(GPU)上进行Kárman漩涡以及叶轮机械叶片尾流等算例仿真,试验表明SMAC法具有更强的视觉真实感和物理真实感,是纯CFD技术的移植和加速,因而具有一定的工程价值;所采用的噪声纹理混合技术能够准确反映流体流动的物理细节,适用于计算机图形学领域,并可应用于矢量场的数据可视化.Stable Fluids provide a standard technique for the simulation of fluid dynamics in computer graphics, but inherently create certain degree of numerical dissipativity. To overcome this shortcoming, we introduce simplified marker and cell （SMAC） method in computational fluid dynamics （CFD） into the computer graphics to perform two-dimensional dynamic simulation of fluids in conjunction with boundary conditions and related techniques used in CFD. Using Stable Fluids and SMAC respectively, we conducted computer simulations on Kdrman vorticity and trailing flow for hydrofoils of turbine machine in engineering situations, running on GPU hardware platform. The results showed that the SMAC based simulation technique displays stronger photo reality and physical reality than one based on Stable Fluids. As a result, it can be used to perform reproducible experiments in engineering applications. Moreover the SMAC noise texture mixing technique can more accurately reflect the physical details of fluid＇s flowing, which is useful in both computer graphics and visualization of data of vector fields.

关 键 词：流体流动仿真 实时 可编程图形硬件 Stable FLUIDS SMAC NAVIER-STOKES方程 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]