基于PVM的三维矢量场流面构造

Parallel Computing of Stream Surface on PVM for 3D CFD Vector Field

文章提出一种三维矢量场自适应粒子跟踪快速算法和基于ＰＶＭ的流面并行计算策略。该算法采用改进的单元搜索方法，消除了原算法计算精度不高和流动特征容易丢失的缺陷。其自适应步长技术，在保证计算精度的前提下，大大减少了流线上的点数，提高了流面的生成速度。实践证明，这种基于ＰＶＭ的并行计算策略进一步加快了算法速度。

A fast algorithm for stream surface for 3D CFD vector field based on particle tracing method is described in this article with cell searching method using tetrahedral decomposition. An adaptive step is used withthe step decreased if the velocity direction changes too much betWeen successive steps, and increased if it changessufficiently little. This greatly reduces the number of points on stream lines without loss of calculating precision. Asimplified triangulation method is use to obtain the stream surface from a set of stream lines. The fast algorithm isimplemented on PVM(Parallel Virtual Machine) to construct its stream surface, which further reduces the calculatingtime.