三维CFD矢量场自适应流线耙并行计算

Parallelization of an adaptive step particle tracingmethod on PVM

在原流线算法的基础上，针对其存在的问题提出了一种新的快速算法．该算法采用四面体分解的单元搜索方法，消除了原算法计算精度不高和流动特征容易丢失的缺陷．根据速度方向的变化来确定粒子跟踪法的积分步长，既保证流线的计算精度，又大大减少了流线上的点数．同时提出了一种基于ＰＶＭ 的并行算法，从而进一步提高了流线耙的生成速度，使流线耙的实时显示成为可能．

A fast algorithm for rakes based on the particle tracing method is described with a new cell searching method using tetrahedral decomposition. This algorithm increases the precision of the stream line and avoids the loss of flow character. When solving the differential equation an adaptive pace is used with the number of steps decreased if the velocity direction changes too much between sucessive steps, and increased if it changes sufficiently little. This greatly reduces the number of points on stream lines without loss of calculating precision. The fast algorithm uses PVM (Parallel Virtual Machine) based on parallel technology to obtain a rake, which further reduces the calculating time at a low cost.