基于GPU的多维树搜索算法在管道流体仿真中的应用

Application of GPU-based Multidimensional Tree Search Algorithm in Pipeline Fluid Simulation

为提高邻近粒子搜索速度,使水力压裂过程中注入管道内的水流现象更好地满足实时性要求,在图形处理器(GPU)上基于光滑粒子流体动力学(SPH)算法,将结合GPU的并行操作能力和多维树搜索算法优点的多维树搜索算法应用在水力压裂管道流体模拟中。首先,给出了流体粒子的受力分析及更新颗粒位置的方法;然后,给出了该算法在管道流体模拟中的具体实现过程。最后,通过实验近似模拟水力压裂过程中管道与水流交互过程,并与其他算法进行对比分析。结果表明将本文算法应用于大规模管道流体模拟中,执行速度快了几倍,实现了水流模拟在管道中的实时性。

In order to improve the searching speed of adjacent particles and make the flow phenomena in the injection pipeline better meet the real-time requirements in the process of hydraulic fracturing,based on the smooth particle hydrodynamics (SPH) algorithm on the graphical processor (GPU),a multi-dimensional tree search algorithm combining the parallel operation ability of GPU and the advantages of multi-dimensional tree search algorithm is applied to the fluid simulation of hydraulic fracturing pipeline.First,the force analysis of fluid particles and the method of updating particle position are given.Then,the implementation process of the algorithm in pipeline fluid simulation is given.Finally,the interaction process between pipeline and water flow during hydraulic fracturing is simulated by experiments,and comparing with other algorithms.The results show that the algorithm can be applied to large-scale pipeline fluid simulation several times faster,and realizes the real-time performance of flow simulation in pipeline.