三维边界元法高阶元几乎奇异积分半解析法

分析了三维边界元法高阶曲面单元几何特征,定义接近度来表征源点与积分单元的接近程度.利用源点在积分单元上的垂足点建立局部极坐标系,构造与几乎奇异积分核函数具有相同奇异性的近似函数.从奇异积分核函数中扣除其近似函数,分离出积分核中主导的奇异函数部分,将奇异积分分解为规则核函数和奇异核函数两项积分.规则核函数积分应用常规Gauss数值积分计算,奇异核函数积分在局部极坐标系ρθ下分离积分变量ρ和θ,对ρ积分建立解析计算列式,对θ积分应用常规Gauss数值积分计算,从而对三维位势问题高阶边界单元几乎强奇异和几乎超奇异积分建立一种新的半解析算法.给出了若干温度场算例,采用边界元法高阶单元几乎奇异积分半解析法计算了近边界内点位势和位势梯度,并与线性单元正则化算法计算结果对比,结果证明提出的半解析法计算几乎奇异面积分和薄壁结构更加高效.      

基于平均值定理和点积分方案的自然单元法及其程序实现

自然单元法是一种基于自然邻接点插值求解偏微分方程的无网格数值方法.它使用Voronoi图或Delaunay三角形作为背景积分网格,使用几何测度构造插值点形函数并形成刚度矩阵.平均值定理定义在未知函数定义域内任何球心(或圆心)的值等于球面(或圆周)上值的平均或加权平均,对于未知函数所满足的平衡方程是充分必要的.因此用平均值定理和点积分方案将求解域内平均应变值由散度定理转化为区域周界上的环路积分,改进传统的积分格式.算例表明,这一积分方案能进一步精简计算量和提高计算效率,是一种自适应的数值计算方法.     

A NEW SEMI-ANALYTIC ALGORITHM OF NEARLY SINGULAR INTEGRALS IN HIGH ORDER BOUNDARY ELEMENT ANALYSIS OF 2D ELASTICITY

针对边界元法中高阶单元中几乎奇异积分计算难题,解剖了二维边界元法高阶单元的几何特征,定义源点相对高阶单元的接近度。将高阶单元上奇异积分核函数用近似奇异函数逼近,从而分离出积分核中主导的奇异函数部分,其奇异积分核分解为规则核函 数和奇异核函数两项积分之和。规则核函数用常规高斯数值积分,再对奇异核函数积分导出解析公式,从而建立了一种新的半解析法,用于高阶边界单元上几乎强奇异和超奇异积分计算。给出3个算例,采用边界元法高阶单元的半解析法计算了弹性力学薄体结构和近边界点位移/应力,并与线性边界元正则化算法结果作了比较,结果表明提出的二次元的半解析算法更加有效。特别是分析薄体结构,采用正则化算法的线性边界元分析比有限元有显著优势,而用提出的二次边界元半解析算法分析比其线性元的有效接近度又减小了4个量级。     

A NEW SEMI-ANALYTIC ALGORITHM OF NEARLY SINGULAR INTEGRALS IN HIGH ORDER BOUNDARY ELEMENT ANALYSIS OF 3D POTENTIAL

分析了三维边界元法高阶曲面单元几何特征,定义接近度来表征源点与积分单元的接近程度.利用源点在积分单元上的垂足点建立局部极坐标系,构造与几乎奇异积分核函数具有相同奇异性的近似函数.从奇异积分核函数中扣除其近似函数,分离出积分核中主导的奇异函数部分,将奇异积分分解为规则核函数和奇异核函数两项积分.规则核函数积分应用常规Gauss数值积分计算,奇异核函数积分在局部极坐标系ρθ下分离积分变量ρ和θ,对ρ积分建立解析计算列式,对θ积分应用常规Gauss数值积分计算,从而对三维位势问题高阶边界单元几乎强奇异和几乎超奇异积分建立一种新的半解析算法.给出了若干温度场算例,采用边界元法高阶单元几乎奇异积分半解析法计算了近边界内点位势和位势梯度,并与线性单元正则化算法计算结果对比,结果证明提出的半解析法计算几乎奇异面积分和薄壁结构更加高效.     

埃尔米特梁单元的分块对角与高阶质量矩阵

埃尔米特梁单元常用的集中质量矩阵,是由挠度自由度对应的一致质量矩阵元素通过行求和或节点积分构造。然而,数值结果表明该集中质量矩阵在求解包含自由端的梁振动问题时,会出现频率精度掉阶现象。本文首先从保障质量矩阵最优收敛性的数值积分精度出发,分别针对三次和五次梁单元,发展了质量矩阵的梯度增强节点积分方案。利用梯度增强节点积分方案,可以得到具有分块对角形式的单元质量矩阵,而其组装的整体质量矩阵除边界节点外仍然呈现对角形式。对于两种单元,其分块对角质量矩阵分别具有4阶最优精度和6阶次优精度。再者,将标准一致质量矩阵和具有同阶精度的梯度增强节点积分质量矩阵进行优化组合,建立了具有超收敛特性的高阶质量矩阵。最后,通过数值算例系统验证了三次和五次单元的分块对角与高阶质量矩阵的频率计算精度。     

边界元法中计算几乎奇异积分的一种无奇异算法

边界元法中存在几乎奇异积分的计算困难。引起边界单元上几乎奇异积分的因素是源点到其邻近单元的最小距离δ。本文拓展文[1]的思想,进一步采用分部积分将δ移出奇异积分式中积分核之外,转换后积分核是δ的正则函数。所以几乎强奇异和超奇异积分被化为无奇异的规则积分与解析积分的和,可由通常的Gauss数值积分解。文中应用此正则化技术求解了弹性力学平面问题的近边界点位移和应力。     

二维位势边界元法高阶单元几乎奇异积分半解析算法

准确计算几乎奇异积分是边界元法难题之一。目前,对于一般的高阶单元的几乎奇异积分尚缺乏通用高效的计算方法。本文在单元局部坐标系中表征了二维高阶单元的几何特征,提出了源点相对高阶单元的接近度概念。针对二维位势边界元法的3节点二次等参单元,构造出与单元积分核具有相同几乎奇异性的近似奇异核函数。从二维位势几乎奇异积分单元积分核中扣除近似奇异核函数,把几乎奇异积分项转换为规则积分和奇异积分两部分之和,规则积分部分用常规Gauss数值积分计算,奇异积分部分由导出的解析公式计算,从而建立了二维位势问题高阶单元几乎强奇异和超奇异积分的半解析算法。算例结果表明了本文半解析算法的有效性和计算精度。     

基于Voronoi结构的无网格局部Petrov-Galerkin方法

基于自然邻结点近似位移函数提出了一种用于求解弹性力学平面问题的无网格局部局部Petrov-Galerkin方法。这种方法在结构求解域Ω内任意布置离散的结点，并且利用需求结点的自然邻结点和Voronoi结构来构造整腐朽 求解的近似位移函数，对于构造好的近似位移函数，在局部Petrov-Galerkin方法建立整体求解的平控制方程，这样平衡方程的积分可在背景三角积分网格的形心上解析计算得到，而采用标准Galerkin方法的自然单元法需要三个数值积分点。该方法能够准确地施加边界条件，得到的系统矩阵是带状稀疏矩阵，对软件用户来说，这它学是一种安全的，真正的无网格方法，所得计算结果表明，该方法的计算精度与有限元四边界单元相当，但计算和形成系统平衡方程的时间比有限元法四边界单元提高了将近一倍，是一种理想的数值求解方法。     

Chebyshev谱元方法结合并行算法求解三维区域的Helmholtz方程

本文探讨了采用Chebyshev谱元方法结合并行计算求解三维区域的Helmholtz方程问题。首先应用变分方法,得到了带有第一类边界条件的三维区域Helmholtz方程的弱形式。然后在三维的标准单元内,采用Chebyshev正交多项式展开函数u和试函数v,并且将其带入弱形式方程,通过积分,得到单元刚度矩阵;通过合成单元刚度矩阵,得到总体矩阵。最后通过基于MPI的并行计算,求解了以总体矩阵为系数的方程组,得到了Helmholtz方程的数值解,和解析解对比表明了数值解的正确性,并且数值解具有8阶精度。在并行求解方程组过程中,充分利用矩阵的对称性和矢量存储来获取上三角元素,这大幅的节约了存储量和计算进程间的通讯量,获得的并行效率可达76.6%。     

有限元方法中的光滑积分伪弱形式

提出了一种光滑积分伪弱形式,将光滑积分拓展至被积函数非偏导项求解。结合光滑应变技术和伪弱形式,可实现有限元系统方程统一光滑积分求解,即对刚度矩阵和质量矩阵中的应变矩阵和形函数矩阵均可进行光滑积分处理,并转化为光滑子域的边界积分。光滑积分伪弱形式与光滑应变技术比较,增加了形函数矩阵不定积分处理过程,且没有降低有限元求解对形函数连续性的要求。不过,伪弱形式改变了单元积分的求解形式,连续质量矩阵求解也无需坐标映射和雅可比矩阵计算。以轴对称二维问题为研究对象,结果表明极度不规则三角形和四边形单元光滑积分伪弱形式在静态和动态有限元方程求解中也具有很好的精度。     

PARTITION OF UNITY FINITE ELEMENT METHOD FOR SHORT WAVE PROPAGATION IN SOLIDS

IntroductionIt is known that standard finite element procedure is unable to simulate the wavepropagation with high oscillations or gradients in space in the media with reasonableefficiency and accuracy due to the nature of polynomial interpolation approxi…     

多边形有限单元形函数的比较研究

多边形有限单元形函数有wachspress插值、Laplace插值和平均值插值三种类型.本文对三种多边形有限单元形函数的性质作了比较研究,给出了三种形函数各自的优点和局限性.Waclaspress和Laplace形函数是有理函数形式,而平均值形函数是无理函数形式.三种形函数均满足单位分解性、线性完备性,且在单元边界上呈线性.在三角形单元上,它们都等价于三角形面积坐标插值.在矩形单元上,Wachspress和Laplace形函数等价于双线性多项式插值形函数.Wachspress和平均值形函数适用于任意凸多边形单元,Laplace形函数更适用于圆内接多边形单元.Wachspress形函数不能推广到含有边节点的单元,平均值形函数可以直接推广到含有边节点的单元.数值试验,验证了本文理论分析的结论.     

A mixed‐interpolation finite element method for incompressible thermal flows of electrically conducting fluids

A new mixed‐interpolation finite element method is presented for the two‐dimensional numerical simulation of incompressible magnetohydrodynamic (MHD) flows which involve convective heat transfer. The proposed method applies the nodal shape functions, which are locally defined in nine‐node elements, for the discretization of the Navier–Stokes and energy equations, and the vector shape functions, which are locally defined in four‐node elements, for the discretization of the electromagnetic field equations. The use of the vector shape functions allows the solenoidal condition on the magnetic field to be automatically satisfied in each four‐node element. In addition, efficient approximation procedures for the calculation of the integrals in the discretized equations are adopted to achieve high‐speed computation. With the use of the proposed numerical scheme, MHD channel flow and MHD natural convection under a constant applied magnetic field are simulated at different Hartmann numbers. The accuracy and robustness of the method are verified through these numerical tests in which both undistorted and distorted meshes are employed for comparison of numerical solutions. Furthermore, it is shown that the calculation speed for the proposed scheme is much higher compared with that for a conventional numerical integration scheme under the condition of almost the same memory consumption. Copyright © 2016 John Wiley & Sons, Ltd.     

基于单位分解法的无网格数值流形方法

在数值流形方法和单位分解法的基础上，提出了无网格数值流形方法. 无网格数值流形 方法在分析时采用了双重覆盖系统，即数学覆盖和物理覆盖. 数学覆盖提供的节点形成求解 域的有限覆盖和单位分解函数；而物理覆盖描述问题的几何区域及其域内不连续性. 与原有 的数值流形方法相比，无网格数值流形方法的数学覆盖形状更加灵活，可以用一系列节点的 影响域来建立数学覆盖和单位分解函数，具有无网格方法的特性，从而摆脱了传统的数值流 形方法中网格所带来的困难. 与无网格方法相比，由于采用了有限覆盖技术，试函数的构造 不受域内不连续的影响，克服了原有的无网格方法在处理不连续问题时所遇到的困难. 详细推导了无网格数值流形方法的试函数和求解方程，最后给出了算例，验证了该方法的正 确性.     

弹性力学的复变量数值流形方法

数值流形方法通过引入数学和物理双重网格,将插值域和积分域分别定义在两个不同的覆盖上来完成系统能量泛函积分运算. 当采用高阶函数构造位移函数时,广义节点自由度将大大增加. 在求解系统的平衡方程中,运算量是与自由度的三次方成正比的,因此数值流形方法的计算量是较大的. 为此,在复变量理论的基础上,采用一维基函数建立二维问题的逼近试函数,然后将其应用于弹性力学的数值流形方法,提出了复变量数值流形方法,推导了弹性力学的复变量数值流形方法的公式. 与传统的数值流形方法相比,复变量数值流形方法具有计算量小、精度高的优点.      

Enriched meshless manifold method for two-dimensional crack modeling

The meshless manifold method is based on the partition of unity method and the finite cover approximation theory which provides a unified framework for solving problems dealing with both continuum with and without discontinuities. The meshless manifold method employs two cover systems. The mathematical cover system provides the nodes for forming finite covers of the solution domain and the partition of unity functions. And the physical cover system describes geometry of the domain and the discontinuous surfaces in the domain. The shape functions are derived by the partition of unity and the finite covers approximation theory. In meshless manifold method, the mathematical finite cover approximation theory is used to model cracks that lead to interior discontinuities in the displacement. Therefore, the discontinuity is treated mathematically instead of empirically by the existing methods. However, one cover of a node is divided into two irregular sub-covers when the meshless manifold method is used to model the discontinuity. As a result, the method sometimes causes numerical errors at the tip of a crack. To improve the precision of the meshless manifold method, the enriched methods are introduced in this work for crack problems.     

Acoustic simulation using a novel approach for reducing dispersion error

It is well‐known that the traditional finite element method (FEM) fails to provide accurate results to the Helmholtz equation with the increase of wave number because of the ‘pollution error’ caused by numerical dispersion. In order to overcome this deficiency, a gradient‐weighted finite element method (GW‐FEM) that combines Shepard interpolation and linear shape functions is proposed in this work. Three‐node triangular and four‐node tetrahedral elements that can be generated automatically are first used to discretize the problem domain in 2D and 3D spaces, respectively. For each independent element, a compacted support domain is then formed based on the element itself and its adjacent elements sharing common edges (or faces). With the aid of Shepard interpolation, a weighted acoustic gradient field is then formulated, which will be further used to construct the discretized system equations through the generalized Galerkin weak form. Numerical examples demonstrate that the present algorithm can significantly reduces the dispersion error in computational acoustics. Copyright © 2016 John Wiley & Sons, Ltd.     

A novel virtual node method for polygonal elements

A novel polygonal finite element method （PFEM） based on partition of unity is proposed, termed the virtual node method （VNM）. To test the performance of the present method, numerical examples are given for solid mechanics problems. With a polynomial form, the VNM achieves better results than those of traditional PFEMs, including the Wachspress method and the mean value method in standard patch tests. Compared with the standard triangular FEM, the VNM can achieve better accuracy. With the ability to construct shape functions on polygonal elements, the VNM provides greater flexibility in mesh generation. Therefore, several fracture problems are studied to demonstrate the potential implementation. With the advantage of the VNM, the convenient refinement and remeshing strategy are applied.     

P-version hybrid analytical finite element method for plate bending problems

Two sets of trial functions with different variables are constructed for the admissible space of the finite element analysis. The trial functions satisfy the equilibrium differential equation inside elements, while the deflections and rotations on the edges of the elements are approximated by the Peano hierarchical interpolation functions. Then, a generalized variational principle is applied to set up the p-version hybrid analytical finite element method for plate bending problems. The accuracy of finite element computation can be improved by increasing the order of the interpolation polynomials with fixed mesh. In the finite element formulation, to obtain the stiffness matrices and the load vectors, it is only necessary to perform quadrature over the edges of the elements. These matrices and vectors possess an embedding structure. The conformability between the elements can be controlled automatically.This work is supported by the Natural Science Foundation of China and the Aeronautical Science Foundation of China.