透過您的圖書館登入
IP:18.191.181.231
  • 期刊

A GPU-Version Lattice Boltzmann Method for Solving Fluid-Particle Interaction Problems

一個GPU版的晶格波茲曼法模擬流體與顆粒交互作用的問題

摘要


發展一個使用GPU(繪圖處理器)加速之沉浸邊界-晶格波茲曼法,模擬流體與顆粒交互作用的問題。本方法使用晶格波茲曼法解不可壓縮流場與沉浸邊界去處理流體與顆粒的交互作用。在沉浸邊界法,推導一個直接力量法去捕捉顆粒的運動。為了印證此數值方法之效率與能力,模擬一個至多個圓球形的顆粒在封閉盒中之沉積作用。數值模擬的結果,相較於CPU,一個球體的沉積作用有30~45倍的效率,很多個球體有40~60倍的效能。

關鍵字

無資料

並列摘要


An immersed-boundary lattice Boltzmann method (IB-LBM) is developed on a graphical processing unit (GPU) to simulate fluid-particle interaction problems. This method uses the lattice Boltzmann method to solve the incompressible flow field and the immersed boundary method to handle the fluid-particle interaction. A direct forcing method is introduced in the IB method to capture the particle motion. To demonstrate of the efficiency and capabilities of the numerical method, a flow in a square pipe and sedimentations of one and many spherical particles in an enclosure are simulated. The numerical results indicate that a speedup of over 60 times when compared to the time required by a single CPU on the flow in a square pipe, a speedup of over 30~45 times on the sedimentation of one spherical particle, and a speedup of about 40~60 times on the sedimentation of large numbers of spherical particles are achieved.

延伸閱讀