等步长高次多项式插值加速度法的局限性

动力时程分析中,在几个相邻的等长时间步之间对加速度的变化规律用多项式插值来描叙,经过推导可求解得到整个时间域上的动力方程的解答.根据泰勒展开原理分析表明,随着所取多项式次数的增加,收敛精度增高,计算步长适当放大,截断误差仍能在容许的范围之内.但是随着所取多项式次数增大,其算法的稳定域减少, 计算步长受到了此小稳定域的限制,收敛精度不再是所取计算步长宽度的决定因素,稳定域大小成了所取计算步长宽度的决定因素.因为一旦步长超出了此小稳定域范围,虽然在每个时间步内的截断误差不大,其传递的误差却会被放大到很多倍,最后导致计算结果严重失真.分析结果显示,多项式插值次数采用到步长的三次时,与一次多项式插值(对应线性加速度法)和二次多项式插值(对应二次加速度法)的分析方法相比,算法的稳定域急剧变窄,为h/T≤0.0099(h为计算步长,T为结构的固有周期),此小稳定域限制了计算步长的选择范围,其收敛精度很高因此可放大计算步长的优势无法施展.本文推导了三次加速度法的求解过程,进行了一个理想单自由度系统的动力时程分析计算,验证了结论的正确性.表明同时考虑收敛精度和稳定域来确定计算步长的宽度时,二次加速度法为优.     

高黏性流动有限元模拟的迭代稳定分步算法

分步算法已被广泛应用于数值求解不可压缩N-S方程. Guermond等认为时间步长必须大于 某个临界值方能使算法稳定. 然而在高黏性流动模拟中，已有的显式和半隐式分步算法由于 其显式本质，必须采用小时间步长计算，不但降低了计算效率，同时也常与为使分步算法稳 分步算法已被广泛应用于数值求解不可压缩N-S方程. Guermond等认为时间步长必须大于 某个临界值方能使算法稳定. 然而在高黏性流动模拟中，已有的显式和半隐式分步算法由于 其显式本质，必须采用小时间步长计算，不但降低了计算效率，同时也常与为使分步算法稳 定必须满足的最小时间步长要求冲突. 本文目的是构造一种含迭代格式的分步算法，它能在 保证精度的前提下大幅度地增大时间步长. 方腔流和平面Poisseuille流数值计算结果证实 了此特点，该方法被有效应用于充填流动过程的数值模拟.     

变截面梁横向振动固有频率数值计算

根据边界条件对变截面梁横向振动四阶变系数微分方程降阶, 形成关于挠度和弯矩的二 阶非显式递推变系数微分方程组; 利用有限差分法, 研究了变截面简支梁横向振动固有频率 的数值计算方法及其精度. 理论分析和正交计算的算例表明: 数值计算算法简单, 计算精度 取决于计算步长的数目和梁横截面竖向渐变率, 与梁宽和梁长无关; 对于给定的计算步长或 数目, 可以估算数值计算的精度; 对于给定的精度要求, 可以确定合理的计算步长或数目.     

应用POD和Padé拟合的线性动力学系统辨识

提出了一种线性动力学系统辨识方法。应用Padé多项式对系统的动刚度曲线进行拟合,通过最小二乘法确定Padé多项式中的系数矩阵,利用遗传算法确定出Padé拟合式中的参数。通过比较系统动刚度矩阵的理论公式和Padé拟合式,辨识得到系统的质量阵、刚度阵和阻尼阵。当系统阶数较高时,可结合POD降阶技术对系统进行辨识,该方法适用于全阶模型和降阶模型。数值仿真算例表明,本文方法具有较高的精度和较好的鲁棒性。     

非线性动力方程精细积分法的自适应步长研究

基于Adams显式和隐式预估公式实现对时间步长的 自适应选择,利用当前时刻v(tk),采用预估公式的两种形式(显式与隐式),对v(tk+1)进行两次预估,利用两公式局部截断误差关系,得出误差估计值ξ(tk+1),并根据其大小 自适应调节时间步长.将该思想应用于预估型(求解过程需要用到预估公式)精细积分算法中,使精细积分算法的时间步长依赖于给定的每步误差限值,提高计算精度,且使算法具有很好的稳定性,对刚度硬化和软化问题均有很好的效果.数值算例验证了本文思想的有效性与适用性.     

结构动力响应分析的三阶显隐式时程积分方法

基于泰勒级数展开式提出了一种用于结构动力响应分析的高精度时程积分方法,该方法假设t时刻的速度和加速度由t-Δt时刻、t时刻、t+Δt时刻的速度和加速度加权表示,并可根据求解需要调节权值,将积分算法构造成隐式格式或显式格式。通过理论分析和数值算例,计算讨论了该算法的稳定性和精度,确定了最佳的权值和允许的时间步长。结果表明:本文算法最高具有三阶精度,且具有振幅衰减率低、周期延长率极小等优点。最后结合一个铁道工程实例,表明本文算法适用于大型非线性动态响应的精确快速求解。     

结构非线性动力方程的精细积分算法

基于线性方程精细积分的思路,对具有惯性、阻尼、刚度非线性的动力方程及参变非线性动力方程提出了一种较高精度线性化精细积分迭代计算算法,算例表明该算法可用较大的步长取得满意的计算精度,并可在较大的线性化区间获得较高的计算精度。     

一种基于Associated Hermite正交函数求解对流扩散方程的算法

提出了一种基于AH(Associated Hermite)正交基函数求解对流扩散方程的无条件稳定算法。该算法将方程的时间项通过Hermite多项式作为正交基函数进行展开,利用Galerkin方法消除时间变量项,从而导出有限维AH域隐式差分方程,突破了传统显式差分格式稳定性条件的限制,最后通过对AH域展开系数的求解得到该对流扩散方程的数值解。在数值算例中,将该算法与传统显示差分法和交替方向隐式差分法进行对比分析,数值计算结果表明,算法无条件稳定且其计算精度与时间步长无关,对于具有精细结构的对流换热问题,该算法具有明显的效率优势,且保持了较高的精度。     

非线性结构动力分析的改进的θ法

本文对结构振动方程的逐步积分法提出了一个二级近似加速度一步法。本方法在有物理阻尼(不论大小)的情况下,是无条件稳定的,具有三阶精度,较好的人工阻尼性质,振幅衰减和周期延长对低振型影响极小,对高振型影响很大。没有超越现象。本方法对解结构动力反应问题,只当刚度。阻尼或步长改变时,每前进一步需解—2n阶线性代数方程;而当刚度、阻尼和步长皆不变时,每前进一步只需完成相当于回代求解的过程即可。比现有的逐步积分方法有明显的优点。     

拟协调等腰梯形壳元显式几何刚度阵及屈曲分析

采用拟协调元方法推导了等腰梯形薄壳元的显式几何刚度阵，用于组合结构屈曲分析的计算，结果表明，这种单元的几何刚度阵收敛快、精度好。     

变截面Timoshenko梁的单元刚度矩阵

变截面构件在工程中应用广泛,在对变截面梁进行数值计算时,需要建立变截面梁单元的刚度矩阵。该文采用势能驻值原理,考虑了轴力引起的几何非线性和剪切变形的影响,将梁截面刚度的变化率作为小量,得到了近似到二阶的单元刚度矩阵。在构造位移模式时,从梁的微分平衡方程出发,得到同样近似到二阶、分别以三次和五次多项式表示的剪切和弯曲位移模式。该文还证明了单元刚度矩阵的奇异性,给出了轴压刚度的表达式,定量论证了与某些精确解的误差,表明在一定范围内,该文的结果具有足够的精度。最后以一个计算实例说明该文的单元刚度矩阵具有较快的收敛性。     

Research on solid-liquid coupling dynamics of pipe conveying fluid

I.IntroductionSolid-fluidcouplingvibrationproblemofpipesconveyingfluidarepresencegenerallyinthedomainofastronomic,energysources,chemicalindustryetc..Notonlytheoreticallytheproblemhaswideresearchvalue,butpracticallytheproblemhaswideengineeringbackground.Therefore,itisimportantreseachproblemihsciencedomainspang.Thefirstrightequationofsolid-liquidcouplingvibrationofpipeconveyingfluidwaspiovidedbyG.W.Housner,andV.Y.Feodosievil'2].Thebasicfrequencycharacteristicofpipesconveyingfluidwasstudiedre…     

Recursive geometric integrators for wave propagation in a functionally graded multilayered elastic medium

The differential equations governing transfer and stiffness matrices and acoustic impedance for a functionally graded generally anisotropic magneto-electro-elastic medium have been obtained. It is shown that the transfer matrix satisfies a linear 1st order matrix differential equation, while the stiffness matrix satisfies a nonlinear Riccati equation. For a thin nonhomogeneous layer, approximate solutions with different levels of accuracy have been formulated in the form of a transfer matrix using a geometrical integration in the form of a Magnus expansion. This integration method preserves qualitative features of the exact solution of the differential equation, in particular energy conservation. The wave propagation solution for a thick layer or a multilayered structure of inhomogeneous layers is obtained recursively from the thin layer solutions. Since the transfer matrix solution becomes computationally unstable with increase of frequency or layer thickness, we reformulate the solution in the form of a stable stiffness-matrix solution which is obtained from the relation of the stiffness matrices to the transfer matrices. Using an efficient recursive algorithm, the stiffness matrices of the thin nonhomogeneous layer are combined to obtain the total stiffness matrix for an arbitrary functionally graded multilayered system. It is shown that the round-off error for the stiffness-matrix recursive algorithm is higher than that for the transfer matrices. To optimize the recursive procedure, a computationally stable hybrid method is proposed which first starts the recursive computation with the transfer matrices and then, as the thickness increases, transits to the stiffness matrix recursive algorithm. Numerical results show this solution to be stable and efficient. As an application example, we calculate the surface wave velocity dispersion for a functionally graded coating on a semispace.     

谱元法和高阶时间分裂法求解方腔顶盖驱动流

详细推导了谱元方法的具体计算公式和时间分裂法的具体计算过程 ;对一般的时间分裂法进行了改进 ,即对非线性步分别用 3阶 Adams-Bashforth方法和 4阶显式 Runge-Kutta法 ,粘性步采用 3阶隐式 Adams-Moulton形式 ,提高了时间方向的离散精度 ,同时还改进了压力边界条件 ,采用 3阶的压力边界条件 ;利用改进的时间分裂方法分解不可压缩 Navier-Stokes方程 ,并结合谱元法计算了移动顶盖方腔驱动流 ,提高了方法可以计算的 Re数 ,缩短了达到收敛的时间 ,并将结果与基准解进行比较 ;分析了移动顶盖方腔驱动流中 Re数对流场分布的影响。     

Notes on the finite element analysis of the axisymmetric elastic solid

A simple transformation of displacements considerably eases the explicit derivation of the finite element stiffness matrix for the axisymmetric elastic solid without causing a decline in the rate of convergence. The worsening of the condition of the global stiffness matrix caused by this transformation can be cured by scaling. A balanced numerical integration scheme maintaining the full rate of convergence is the one that integrates each term of the work and energy expressions to the order 2p ? 2, p being the degree of the complete polynomial in the shape functions.     

Sparse matrix implicit numerical integration of the Stiff differential/algebraic equations: Implementation

The finite element absolute nodal coordinate formulation (ANCF) is often used in modeling very flexible bodies in multibody system (MBS) applications. This formulation leads to a constant mass matrix, allowing for an efficient sparse matrix implementation. Nonetheless, the use of the ANCF finite elements to model stiff structures can lead to high frequencies associated with ANCF coupled deformation modes, as discussed in the literature. Implicit numerical integration methods can be effectively used to develop efficient procedures for the solution of MBS differential/algebraic equations. Most existing implicit integration algorithms, however, require numerical differentiation of the equations of motion, and some of these integration methods do not ensure that the kinematic algebraic constraint equations are satisfied at all levels (position, velocity, and acceleration). Because of these limitations, existing implicit integration methods can be less accurate and less efficient when used to solve large scale MBS applications. In order to circumvent this problem, the two-loop implicit sparse matrix numerical integration (TLISMNI) method was proposed for the solution of MBS differential/algebraic equations. The TLISMNI method does not require numerical differentiation of the forces and allows for an efficient sparse matrix implementation. This paper discusses TLISMNI implementation issues including the step size selection, the error control, and the effect of the numerical damping. The relation between the step size selection and the structure stiffness is also discussed. The use of the computer implementation described in this paper is demonstrated by solving very stiff structure problems using the Hilber?CHughes?CTaylor (HHT) method, which includes numerical damping. An eigenvalue analysis and Fast Fourier Transform (FFT) are performed in order to identify the fundamental modes of deformation and demonstrate that the contributions of these fundamental modes can be erroneously damped out when some other implicit integration methods are used. The TLISMNI method, on the other hand, captures the contributions of these fundamental modes. The results, obtained using the TLISMNI method, are compared with the results obtained using other methods including the implicit HHT-I3 and the explicit Adams integration methods. The results obtained show that the TLISMNI method can be five times faster than the other two methods when no numerical damping is considered.     

移动荷载作用下非局部地基梁动力响应分析的有限元法Dynamic response of beams with nonlocal foundations subjected to moving loads using finite element method

基于非局部地基理论,推导了移动荷载作用下非局部地基梁动力响应问题的有限元解,分别讨论了地基的非局部参数、刚度、阻尼系数以及移动荷载速度对非局部地基梁动力响应的影响,并比较了非局部结果与局部结果的差异。结果表明,地基的非局部参数、刚度和阻尼是地基梁的动力响应的主要影响参数,地基梁最大响应及其发生的时刻与移动荷载速度有关。研究成果可为轨道地基系统设计提供参考。     

模拟火炮后坐运动的非线性弹簧－阻尼单元

根据非线性有限元法的基本理论，建立了火炮发射动力学的有限元运动方程，推导出单元耗散为向量和单元阻尼矩阵的计算公式。提出采用非线性弹簧－阻尼单元来模拟反后坐装置对炮身后坐的阻尼作用，并给出了该单元的刚度矩阵和阻尼矩阵。根据炮身的后坐过程的三个阶段，提出采用分段拟合的方法确定非线性单元的参数以模拟火炮后坐阻力－位移曲线。采用非线性有限元单元模拟后坐装置，能够较准确地描述炮身部分的后坐运动，同时在大简化了全炮模型的建立，为全炮的动力学仿真提出了一条新的思路。     

Method of model analysis for flexible head impacting with elastic plane

ＩｎｔｒｏｄｕｃｔｉｏｎＲｅｓｅａｒｃｈｅｓｓｈｏｗ :70ｔｏ 80ｐｅｒｃｅｎｔｏｆｔｈｅｖｅｈｉｃｌｅｓｔｈａｔｃａｕｓｅｔｈｅａｃｃｉｄｅｎｔｓａｒｅｓａｌｏｏｎｃａｒｓ[1].Ｉｎｔｈｅ 70ｐｅｒｃｅｎｔｏｆｔｈｅａｃｃｉｄｅｎｔｓ,ｔｈｅｈｕｍａｎｂｏｄｙｗｉｌｌｃｏｌｌｉｄｅｗｉｔｈｔｈｅｆｒｏｎｔｏｆｔｈｅｃａｒａｎｄｔｈｅｈｅａｄａｎｄｌｅｇｗｉｌｌｂｅｉｎｊｕｒｅｄ .Ｍｅａｎｔｉｍｅ ,ｔｈｅｈｏｏｄａｎｄｗｉｎｄｓｈｉｅｌｄｗｉｌｌｃａｕｓｅｔｈｅｉｎｊｕｒｙｏｆｔｈｅｈｅａｄａｎｄｔｈ…