可扩展的冲击—接触并行计算研究

冲击—接触计算模型在汽车碰撞、金属成型等的模拟计算中有着广泛的应用，鉴于冲击—接触计算过程复杂和计算量大，本文在分布式可扩展的并行计算平台上，设计并实现了冲击—接触的并行计算。算例证明，计算平台稳定可靠，算法简单实用，且具有较好的并行效率和可扩展性。     

高阶紧致格式分区并行算法

针对超声速多尺度复杂流动问题,发展了一种高精度并行算法。计算格式采用五阶迎风紧致格式,用特征型通量限制方法抑制非物理振荡。在对接边界处采用五阶WENO格式,以保证整个计算域内计算精度一致。通过网格分区和数据交换,在MPI平台实现了并行计算。通过超声速算例对算法进行了验证,并对并行效率和加速比进行了分析。最后,将算法应用于超声速转捩、湍流问题的数值模拟。计算结果表明,提出的算法具有较高的精度和分辨率,对接边界光滑连续,并且并行效率较高,在高超声速湍流流动数值模拟中取得了较好的应用效果。     

结构非线性动力分析混合时间积分并行算法

综合隐式和显式时间积分技术,对结构非线性动力反应分析提出一种并行混合时间积分算法.该算法采用区域分解技术.将并发性引入到算法中,即利用显式时间积分技术进行界面节点积分而利用隐式算法求解局部子区域.为实现并行混合时间积分算法,设计了灵活的并行数据信息流.编写了该算法的程序,在工作站机群实现了数值算例,验证了算法的精度和性能.计算结果表明该算法具有良好的并行性能,优于隐式算法.     

并行加速的局部变分迭代法及其轨道计算应用

为满足航天工程对轨道计算精度和实时性的高要求,近年来发展出了可以通过大步长积分修正实现快速精确求解的积分修正类方法.积分修正类方法有可并行计算的特点,然而在串行计算环境下会受到计算资源的限制,无法充分发挥其可并行加速的优势.此外,合理的计算参数通常难以预先确定,也使积分修正类方法大步长快速计算的优势难以充分体现.针对以上问题,利用积分修正类方法可并行计算的特点,提出了并行加速的局部变分迭代法PA-LVIM,通过将传统局部变分迭代法LVIM的并行计算量均摊到多个计算节点上,显著提高了计算速度.此外,还使用根据系统状态二阶导数分布确定计算参数的打磨法优化了PA-LVIM的计算参数,进一步发挥了其大步长快速计算的优势.求解了三个经典的轨道递推问题,仿真结果表明, PA-LVIM的加速效果明显,且经打磨法优化计算参数后,其计算效率又进一步得到提高,将当前主流方法的计算效率提高了5倍以上.     

????????????????????????????

合隐式和显式时间积分技术,对结构非线性动力反应分析提出一种并行混合时间积分算 法. 该算法采用区域分解技术. 将并发性引入到算法中,即利用显式时间积分技术进行界面 节点积分而利用隐式算法求解局部子区域. 为实现并行混合时间积分算法,设计了灵活的并 行数据信息流. 编写了该算法的程序,在工作站机群实现了数值算例,验证了算法的精度和 性能. 计算结果表明该算法具有良好的并行性能,优于隐式算法.     

接触-碰撞算法研究进展

接触-碰撞广泛存在于实际工程问题中,是影响数值计算效率与计算精度的重要因素。本文针对变形体间接触-碰撞问题的显式有限元计算,介绍接触-碰撞算法近30年来取得的主要进展。首先,简要介绍接触-碰撞问题的界面离散模型;然后,从全局接触搜索、局部搜索、接触约束施加以及接触计算的并行化四方面详细阐述目前主要算法的基本思想与特点,并分析其优势与不足;最后,对接触-碰撞算法相关研究方向给出建议。     

接触-碰撞算法研究进展

接触-碰撞广泛存在于实际工程问题中,是影响数值计算效率与计算精度的重要因素。本文针对变形体间接触-碰撞问题的显式有限元计算,介绍接触-碰撞算法近30年来取得的主要进展。首先,简要介绍接触-碰撞问题的界面离散模型;然后,从全局接触搜索、局部搜索、接触约束施加以及接触计算的并行化四方面详细阐述目前主要算法的基本思想与特点,并分析其优势与不足;最后,对接触-碰撞算法相关研究方向给出建议。     

弹塑性摩擦接触问题形状设计灵敏度分析

提出了一种计算二维有限变形弹塑性摩擦接触问题形状设计灵敏度的算法. 采用主动集策略和mortar方法处理接触边线上的约束条件. 在mortar接触边线的切线和法线方向上采用相同的名义罚函数,提出基于名义罚函数的移动摩擦锥算法来正则化接触约束条件,发展了一种新的二维多体有限变形摩擦接触算法. 在此基础上, 通过将离散形式的摩擦接触问题控制方程对形状设计变量微分,得到了该路径相关问题的直接微分法解析设计灵敏度计算格式, 其节点位移灵敏度方程在每个增量步不用迭代、直接求解. 与国际上现有的二维多体有限变形摩擦接触问题的解析设计灵敏度算法相比,本算法不需分解为法向和切向推导,表达式较简洁,便于编程实现. 数值算例验证了算法的精度和有效性.      

高雷诺数湍流风场大涡模拟的并行直接求解方法

在环境流体力学中,风场是风沙流、风雪流等自然环境特性问题研究的动力源和基础.通常采用壁湍流模型进行风场大涡模拟(large eddy simulation, LES)计算,但受到计算规模的限制使得高雷诺数风场的模拟计算难以实现.并行计算技术是解决大规模高雷诺数风场大涡模拟的关键技术之一.在不可压湍流风场的LES模拟中,压力泊松方程的并行计算技术是进行规模并行计算的困难点.根据风场流动模拟计算的特点,采用水平网格等距而垂直于地面网格非等距,在解决规模并行计算中求解压力泊松方程的难点问题时,利用FFT解耦三维泊松方程使其变为垂向的一维三对角方程,并利用可并行的三对角方程PDD求解技术,可建立三维泊松方程的直接并行求解技术.结合其它容易并行的动量方程计算,本文建立风场LES模拟的并行直接求解方法 (parallel direct method-LES, PDM-LES).在超级计算机上对新方法进行并行效率测试,并行计算效率达到90%.新的方法可用于进行湍流风场大涡模拟的大规模并行计算.计算结果表明,湍流风场瞬时速度分布近壁面存在条带状的拟序结构,平均场的速度分布符合速度对数律特性,风场湍流特性基本合理.     

岩体渗透结构类型的划分及其渗透特性研究

提出了一种计算二维有限变形弹塑性摩擦接触问题形状设计灵敏度的算法. 采用主动 集策略和mortar方法处理接触边线上的约束条件. 在mortar接触边线的切线和法线方向上 采用相同的名义罚函数,提出基于名义罚函数的移动摩擦锥算法来正则化接触约束条件,发 展了一种新的二维多体有限变形摩擦接触算法. 在此基础上, 通过将离散形式的摩擦接触问题 控制方程对形状设计变量微分,得到了该路径相关问题的直接微分法解析设计灵敏度 计算格式, 其节点位移灵敏度方程在每个增量步不用迭代、直接求解. 与国际上现有 的二维多体有限变形摩擦接触问题的解析设计灵敏度算法相比,本算法不需分 解为法向和切向推导,表达式较简洁,便于编程实现. 数值算例验证了算法的精度 和有效性.     

随机有限元分析的二级区域分解并行求解算法

采用蒙特卡罗模拟(MCS)和加权积分法对二维问题进行随机有限元分析。尽管MCS方法对任何有确定解的问题都具有求解精度高的优点,但由于求解所需的计算量巨大使其应用受到限制。利用并行求解技术可有效地处理这种密集型计算问题。基于有限元分裂对接法(FETI)的并行特性并利用预处理共轭梯度法(PCG)的求解高效性,结合整体子区域实现(GSI-PCG)和FETI法,提出二级求解算法,并在工作站机群上实现了数值算例。算例计算结果表明本文GSI(PCG)-FETI算法具有较高的并行加速比和并行效率,具有良好的性能,可有效地进行二维问题的随机有限元分析。     

一类针对高阶单元接触搜索算法：系统搜索算法

提出一类基于非线笥约束优化理论的高副搜索算法。通用的一些接触搜索算法一般针对低阶单元,它们具有较高的计算效率,但不适应于高阶单元。随着工整中对计算精度要求不断提高,高阶单元逐步被引入到接触问题计算程序中。因此,给出针对亢介单元而人有与现有的低阶算法相比扩计算效率的接触索算法具有现实意义,本文采用系统索方法和非线性约束优化理论,建立一种高阶单元接触点副搜索算法。计算表明,本算法具有较高的计算疚和良好     

高质量点集的快速局部网格生成算法

高效及高质量的局部网格生成算法是基于节点有限元并行方法设计的关键。泡泡布点算法能够在复杂区域上不经过人工干预生成高质量的节点集,本文提出了基于该方法所生成的节点集的快速局部网格生成算法。该算法充分利用泡泡布点方法提供的节点集及节点邻接链表信息,避免了桶数据结构的建立以及节点的局部搜索过程,只需应用Delaunay三角剖分的外接圆准则从中心节点的邻接链表中去除极少数的非卫星点,可快速地生成局部网格,比现有的局部网格生成算法更为快捷。算例结果表明,该算法高效可靠,生成网格与Delaunay三角剖分网格一致。     

Parallel finite element algorithm based on full domain partition for stationary Stokes equations

Based on the full domain partition, a parallel finite element algorithm for the stationary Stokes equations is proposed and analyzed. In this algorithm, each subproblem is defined in the entire domain. Majority of the degrees of freedom are associated with the relevant subdomain. Therefore, it can be solved in parallel with other subproblems using an existing sequential solver without extensive recoding. This allows the algorithm to be implemented easily with low communication costs. Numerical results are given showing the high efficiency of the parallel algorithm.     

Linear and Nonlinear Algorithms for Stress Separation in Photoelasticity

An experimental-numerical hybrid method for the stress separation in photoelasticity is proposed in this study. In the proposed method, boundary conditions for a local finite element model, that is, tractions along boundaries are inversely determined from photoelastic fringes. Two algorithms are proposed for determining the boundary condition. One is a linear algorithm in which the tractions are obtained by the method of linear least-squares from both principal stress difference and principal direction. Another is the nonlinear algorithm in which the tractions are determined only from the principal stress difference. After determining the boundary conditions for the local finite element model, the stresses can be obtained by finite element direct analysis. The effectiveness is demonstrated by applying the proposed method to a perforated plate under tension and contact problems. Results show that the boundary conditions of the local finite element model can be determined from the photoelastic fringes and then the individual stresses can be obtained by the proposed method. Furthermore, the stresses can be evaluated even if the boundary condition is complicated such as at the contact surface. It is expected that the proposed method can be powerful tool for stress analysis.     

Local and parallel finite element algorithms based on two-grid discretization for steady Navier-Stokes equations

Local and parallel finite element algorithms based on two-grid discretization for Navier-Stokes equations in two dimension are presented. Its basis is a coarse finite element space on the global domain and a fine finite element space on the subdomain. The local algorithm consists of finding a solution for a given nonlinear problem in the coarse finite element space and a solution for a linear problem in the fine finite element space, then droping the coarse solution of the region near the boundary. By overlapping domain decomposition, the parallel algorithms are obtained. This paper analyzes the error of these algorithms and gets some error estimates which are better than those of the standard finite element method. The numerical experiments are given too. By analyzing and comparing these results, it is shown that these algorithms are correct and high efficient.     

A parallel two-level finite element method for the Navier-Stokes equations

Based on domain decomposition, a parallel two-level finite element method for the stationary Navier-Stokes equations is proposed and analyzed. The basic idea of the method is first to solve the Navier-Stokes equations on a coarse grid, then to solve the resulted residual equations in parallel on a fine grid. This method has low communication complexity. It can be implemented easily. By local a priori error estimate for finite element discretizations, error bounds of the approximate solution are derived. Numerical results are also given to illustrate the high efficiency of the method.     

An efficient parallel and fully implicit algorithm for the simulation of transient free-surface flows of multimode viscoelastic liquids

The parallelization of a fully implicit and stable finite element algorithm with relative low memory requirements for the accurate simulation of time-dependent, free-surface flows of multimode viscoelastic liquids is presented. It is an extension of our multi-stage sequential solution procedure which is based on the mixed finite element method for the velocity and pressure fields, an elliptic grid generator for the deformation of the mesh, and the discontinuous Galerkin method for the viscoelastic stresses [Dimakopoulos and Tsamopoulos [12], [14]]. Each one of the above subproblems is solved with the Newton–Rapshon technique according to its particular characteristics, while their coupling is achieved through Picard cycles. The physical domain is graphically partitioned into overlapping subdomains. In the process, two different kinds of parallel solvers are used for the solution of the distributed set of flow and mesh equations: a multifrontal, massively parallel direct one (MUMPS) and a hierarchical iterative parallel one (HIPS), while viscoelastic stress components are independently calculated within each finite element. The parallel algorithm retains all the advantages of its sequential predecessor, related with the robustness and the numerical stability for a wide range of levels of viscoelasticity. Moreover, irrespective of the deformation of the physical domain, the mesh partitioning remains invariant throughout the simulation. The solution of the constitutive equations, which constitutes the largest portion of the system of the governing, non-linear equations, is performed in a way that does not need any data exchange among the cluster's nodes. Finally, indicative results from the simulation of an extensionally thinning polymeric solution, demonstrating the efficiency of the algorithm are presented.     

An iterative parallel algorithm of finite element method

In this paper, a parallel algorithm with iterative form for solving finite element equation is presented. Based on the iterative solution of linear algebra equations, the parallel computational steps are introduced in this method. Also by using the weighted residual method and choosing the appropriate weighting functions, the finite element basic form of parallel algorithm is deduced. The program of this algorithm has been realized on the ELXSI-6400 parallel computer of Xi'an Jiaotong University. The computational results show the operational speed will be raised and the CPU time will be cut down effectively. So this method is one kind of effective parallel algorithm for solving the finite element equations of large-scale structures.     

基于PVM的网络并行有限元面向对象的设计

子结构是有限元并行计算常用的一种方法,本文采用面向对象的方法,首先对子结构进行了面向对象的设计,得到了其类层次结构图;然后针对工作站网络有限元并行计算环境。提出了基于PVM消息传递平台上的Shadow—Mirror数据传输模型,该模型在有限元并行计算数据传输时,充分发挥数据面向对象的特性,采用设置数据缓冲区、短消息合并等方法以缩短数据通信时间,并据此编制了相应的程序。计算结果表明,使用文中提出的面向对象的Shadow—Mirror数据传输模型可以得到较为理想的并行加速比,而且随着问题规模增大,并行加速比增高。本文研究内容为进一步开展基于工作站网络的并行有限元研究提供了一个可参考的基础。