【作者】 曹国强；

【导师】 梁冰；

【作者基本信息】 辽宁工程技术大学 ， 工程力学， 2005， 博士

【摘要】 随着现代计算流体力学(CFD)的发展,对高速度、大容量计算机的需求越来越迫切,计算越来越复杂,规模也呈指数级增长。如此庞大的运算任务单台计算机显然无法胜任。一方面是单处理器性能发展空间有限,另一方面是专用高档并行计算机价格又太昂贵,而且旧有的资源又得不到持续和有效的利用,造成硬件资源空闲和资金浪费。因此,一种能够充分利用现有计算机软硬资源,有效发挥计算资源协同增强作用,同时满足不同专业研究人员高效方便处理的并行计算技术、平台和实现方法就变得十分重要和具有现实意义。 本文在综合国内外关于流体力学并行计算及相关领域研究文献的基础上,搭建了微机并行计算平台,解决了数值模拟对计算资源要求;对并行环境下子区域网格的独立生成和相邻子区域交界面的网格生成进行研究;在原有三维欧拉方程求解器的基础上添加粘性(耗散)项,针对定常问题,考虑粘性的影响,加入无滑移边界条件和两方程湍流模型,得到描述湍流运动的封闭RANS方程;对离心式叶片泵的设计理论进行了深入研究,并利用Visual C++平台编写了基于Pro/E的三维参数化设计和计算软件。最后利用以上得出的计算经验和结论,通过对一个实际工程问题—离心式叶片泵内部流动的数值计算,完成了CFD中湍流高级数值模拟技术的尝试。全文共七章。 第一章为绪论,综述了国内外有关并行计算技术的研究概况,阐述了本文的研究目的、意义与主要研究内容。 第二章对Windows2000操作系统下的并行计算环境进行了深入研究,在此基础上成功搭建了由4台PC组成的局域网并行集群系统,采用消息传递接口MPI作为并行程序解决方案,安装了MPICH并行程序运行环境,用C语言调用MPI消息传递库,并在此基础上编写和实现了流场计算中应用到的并行迭代程序、矩阵二维拓扑和三维拓扑并行程序。 第三章研究了并行环境下子区域网格的独立生成和相邻子区域交界面的网格生成。在网格并行生成过程中,给出了关于边的分类,完善了文献中子区域内网格生成时接受新点及新单元的条件,给出了流场整体网格图以及流场子区域、交界面局部网格图。 第四章对κ-ε两方程模型进行了深入研究,在原有三维欧拉方程求解器的基础上添加粘性(耗散)项,针对定常问题,考虑粘性的影响,加入无滑移边界条件和两方程湍流模型,得到描述湍流运动的封闭RANS方程。 第五章利用Fluent软件在搭建的并行计算平台上对离心式叶片泵的内部流场进行数值模拟,对湍流κ-ε两方程模型进行验证,并得到了一些结论。 第六章通过对离心式叶片泵的优秀水力模型进行了深入的研究,在一元和三元流动理论的基础上给出了叶轮设计中的各种参数计算公式;利用Visual C++平台更多还原

【Abstract】 With the development of modern computer fluid dynamics, the need for high performance and high capacity becomes more and more urgent., computation becomes more and more complex, scale develops with positive exponent. For such tremendous calculation tasks, a single computer is not able to complete these computations obviously: one is the limited capability of single processor; the other is those special expensive parallel computers are too expensive and formerly resource have to be abandoned, which is waste. Therefore, it is necessary and realistic meaning for us to develop an effective parallel calculating technology, platform and realizing method for different researchers, at the same time we can use present computer software and hardware resources fully. In the basis of the references on a centrifugal pump of two-phase fluids from domestic and oversea, its theory of boundary layer is presented and set up and the modular curve equation of vane is received in this paper. It divides into 7 chapters.The introduction is in chapter one, which summarizes the research results of parallel computation technology in domestic and oversea and illuminates the research aim, significant and main contents.In chapter two, through deep research to windows 2000 operation system, wehave construct a parallel PC cluster-based system with 4 computers; also can extending more computer nodes easily. We use message passing interface (MPI) as a method to program parallel software, then MPICH has been installed and use C langue to call MPI function warehouse. Finally, we have made and realized parallel iterative arithmetic, matrix of two-dimensional and topological arithmetic, three-dimensional topological arithmetic, molecule dynamics simulation and three-dimensional graphics processing parallel programs.In chapter three, we researches the generation of sub domain mesh and the adjacent domain mesh under parallel environment, firstly improve the R lohner’s wave front domain-splitting algorithm so that the sub grids and their boundaries are more favorable for grid generation. A new optimization strategy of sub domain’s boundary is then presented in order to improve the smoothness of boundaries and the quality of grids. After subdividing the initial meshes by using the above domain-splitting algorithm, we smooth the grid by the parallel method.In chapter four, the NS solver is developed from Euler equation code by adding viscous terms and two-equation turbulence models. In order to get a better result, the improved Jameson second and forth artificial dissipation is used, the implicit residual更多还原