Simulation of Heat Conduction in Fluids on GPU with Particle Method

Darcy Qingzhi Hou,Zhe Wang,Jianwu Dang,Wenhuan Lu,Yuejin Cai,Jianguo Wei
2017-01-01
Abstract:Fluid simulation is an important topic in virtual reality. To accurately simulate heat conduction in flowing fluids, a Lagrangian particle model is proposed in this paper, which is based on the meshless smoothed particle hydrodynamics (SPH) method. For large simulation with massive particles, running SPH on CPU can be very time consuming. To take advantage of the powerful parallel computing, Compute Unified Device Architecture (CUDA) is used to implement the developed particle model on GPU. Numerical results indicated that the SPH method is a good option for simulating heat conduction phenomenon in fluids with high accuracy. Neither numerical diffusion nor spurious oscillation is observed. For simulations with massive particles, the computational speed is significantly improved on GPU, giving a good speedup curve. The parallelized Lagrangian particle model on GPU with CUDA is a promising tool for the simulation of large scale heat transport in fluids
What problem does this paper attempt to address?