摘要
对于笛卡尔网格方法,壁面距离是采用虚拟单元法精确处理物面边界的重要参数,同时也是网格自适应后制约流动计算效率的关键因素之一。针对现有壁面距离计算方法结果不精确、效率不高的问题,引入三角形参数化方法,将空间点到三角形物面离散网格的最小距离问题转换为约束条件下一维极值问题,仅需通过符号判断和少量加减乘运算,即可确定最小距离,计算精度和效率大幅提高;发展嵌套包围盒概念的KDT(K-dimensional tree)物面网格数据存储结构,优化KDT最近邻搜索算法中距物面较远数据点回溯过程,实现了最小距离对应的三角形的快速定位。运用球、导弹、DPW6等三维几何构型对上述方法考核验证结果表明,计算得到的壁面距离与解析值的误差在百万分之一以内,十亿量级网格规模下的单核计算效率接近已有文献中的并行计算效率。
The wall distance of Cartesian grids is an essential parameter for the proper wall treatment using ghost cells and is also one of the critical factors governing the efficiency of the flow field simulation after mesh adaptation.This paper proposes a triangular parameterization method that converts the problem of computing the minimum distance between spatial points and discretized triangular meshes on the surface into a constrained onedimensional extremum problem.This simplification only requires symbolic judgments and a small number of addition,subtraction and multiplication operations to obtain the minimal distance,yielding significant improvements in the accuracy and efficiency compared to existing methods.Meanwhile,a KDT(K-dimensional tree) data structure based on the nested enclosing box concept is developed to optimize the backtracking of data points far from the surface in the KDT nearest neighbor search algorithm.The application of this method to three-dimensional geometries such as spheres,missiles,and DPW6 demonstrates that the error between the computed wall distance and the resolved one is within one millionth.Moreover,the computational costs of using a single core for billion-scale grids are comparable to those of parallel computation using existing methods.
作者
孟爽
周丹
李雪亮
毕林
MENG Shuang;ZHOU Dan;LI Xueliang;BI Lin(Key Laboratory of Traffic Safety on Track(Central South University),Ministry of Education,Changsha 410075,China;State Key Laboratory of Aerodynamics,Mianyang 621000,China;Computational Aerodynamics Institute of China Aerodynamics Research and Development Center,Mianyang 621000,China)
出处
《空气动力学学报》
CSCD
北大核心
2023年第7期93-101,I0002,共10页
Acta Aerodynamica Sinica
基金
国家数值风洞工程(NNW2018-ZT1A02)
中南大学研究生自主探索创新项目(206021722)。
关键词
笛卡尔网格
壁面距离
计算效率
KDT
回溯方法
Cartesian grid
wall distance
computational efficiency
K-dimensional tree
backtracking