一维C~1有限元后处理超收敛计算的新型算法

该文针对一维C~1有限元提出一种新型后处理超收敛算法,由该法可求得全域超收敛的位移和内力。该法在单个单元上逐单元实施,通过将单元端部结点位移有限元解设为本质边界条件,在单元域上建立单元位移恢复的局部边值问题。对该局部边值问题,以单元内任一点为结点将单元划分为两个子单元进行有限元求解,子单元次数与原单元相同,由此获得该点位移的超收敛解。对单元内所有点均作这样的超收敛求解,可获得整个单元上位移的超收敛解。该位移超收敛解光滑、连续,通过对该位移超收敛解求导可获得转角和内力的超收敛解。数值结果表明,对于m次元,该法得到的挠度和转角具备与结点位移相同的h~(2m-2)阶的最佳收敛阶;弯矩和剪力则分别具备h~(2m-3)、h~(2m-4)阶的收敛阶,均比相应有限元解高出m-2阶。该法可靠、高效、易于实施,是一种颇具潜力的后处理超收敛算法。     

二维有限元单元角结点位移精度修正之初探

二维四边形有限元单元角结点位移相较于其他结点位移,有更高的收敛阶。对于足够光滑的问题,采用m次单元,其角结点位移收敛阶最高可达2 m阶。该文以二维Poisson方程为例,在有限元解的基础上,利用单元能量投影(EEP)法的超收敛解计算残余荷载向量,在不改变整体刚度矩阵的基础上,仅需进行代数方程组回代,即可得到具有更高精度的单元角结点位移。数值结果表明:当采用EEP简约格式解计算残余荷载向量时,单元角结点位移收敛阶最高可提高为2m+2阶。特别地,对于线性元,精度翻倍,效益十分显著。     

一维Ritz有限元超收敛计算的EEP法简约格式的误差估计

该文对一维问题Ritz有限元后处理超收敛计算的EEP(单元能量投影)法简约格式给出误差估计的数学证明,即对足够光滑问题的(>1)次单元的有限元解答,采用EEP法简约格式计算得到的单元内任一点位移和应力(导数)超收敛解均可以达到的收敛阶,即位移比常规有限元解的收敛阶至少高一阶,而应力则至少高二阶。     

一维C~1有限元EEP超收敛位移计算简约格式的误差估计

该文对一维C1有限元后处理超收敛计算的EEP(单元能量投影)法简约格式中的位移解给出误差估计的数学证明,即对足够光滑问题的m(3)次单元的有限元解答,采用EEP法简约格式得到的单元内任一点位移超收敛解均可以达到hm+2的收敛阶,比常规有限元位移解的收敛阶至少高一阶。     

平面曲梁有限元静力分析的p型超收敛算法

该文对平面曲梁有限元静力分析提出一种p型超收敛算法,由该法可求得曲梁结构全域超收敛的位移和内力。该法基于有限元解答中结点位移的超收敛特性,通过将单元端部结点位移有限元解设为本质边界条件,在单元上建立单元位移近似满足的线性常微分方程边值问题,对该边值问题采用更高次数的多项式进行有限元求解获得单元上位移的超收敛解,将位移超收敛解代入内力表达式获得内力的超收敛解。该法简单、直接,通过很少量的计算即能显著提高位移和内力的精度和收敛阶。数值结果显示,该法高效、可靠,是一个颇具潜力的方法。     

运动方程一阶方程组格式的线性时域有限元及其EEP超收敛计算

该文将运动方程转换成一阶常微分方程组,采用Galerkin线性单元,构建相应的h 2阶精度的递推公式,并基于单元能量投影(EEP)法进行结点位移修正得到h 4阶精度的有限元结点解。该文中对其稳定性和收敛阶给出数学分析和证明,同时给出了一个自适应步长算法,并通过数值算例验证其不失为一种有效、简洁的时域积分算法。     

运动方程自适应步长求解的高性能Galerkin时程单元初探

该文以一阶运动方程为例,利用其非自伴随性质,构建了新型的凝聚检验函数,进而提出了一套高性能Galerkin有限单元——凝聚单元.该单元为无条件稳定的单步法单元,对于(m)次多项式单元,其端结点位移和速度均可达到O(h2(m)+2)阶的超高收敛性,比常规Galerkin单元的结点精度高2阶.采用此单元,该文进而实现了无需...     

具有最佳超收敛阶的Galerkin有限元EEP法计算格式

对二阶非自伴问题的一维Galerkin有限元法提出其后处理超收敛计算的EEP(单元能量投影)法改进的最佳超收敛计算格式,即用m次单元对足够光滑问题的Galerkin有限元解答,采用该格式计算的任一点的位移和应力都可以达到h2m阶的最佳超收敛结果。该文首先针对高次单元提出了凝聚试探形函数和凝聚检验形函数的概念,证明了相关的逼近定理和等价定理,然后给出了具体的算法公式。最后给出了一系列典型的数值算例用以验证这种最新的EEP法改进格式确实能够使位移和导数逐点达到最佳收敛阶。     

二维泊松方程问题Lagrange型有限元p型超收敛算法

该文针对二维泊松方程问题的Lagrange型有限元法提出了一种p型超收敛算法。该法受有限元线法对二维问题降维思想的启发,基于网格结点位移的天然超收敛性,通过从网格中取出一行对边相邻的单元作一子域,将子域内各单元另一对边解答取为原有限元解答,在子域上建立真解近似满足的局部偏微分方程边值问题,对该局部边值问题,沿对边方向单向提高单元阶次进行有限元求解获得单元对边上的超收敛解。单元另一对边上的超收敛解可通过另一方向的单元行类似获得。在单元边超收敛解的基础上,依次取出各个单元,以单元边位移超收敛解为Dirichlet边界条件,双向提高单元阶次对原泊松方程问题进行有限元求解即可获得全域超收敛解。数值算例表明,通过简单的后处理计算本法可显著提高解答的精度和收敛阶。     

运动方程自适应步长求解的一个新进展——基于EEP超收敛计算的线性有限元法

该文采用最简单的Galerkin型线性单元,对运动方程构建了简捷高效的单步法递推公式;进而基于EEP超收敛计算技术,开发了单元步长自动优化和结点位移精度修正两项关键技术,可在整个时域上得到误差分布均匀且逐点满足给定的误差限的解答——堪称数值解析解。该文给出了单自由度和多自由度的数值算例以验证本法的有效性。     

具有最佳超收敛阶的EEP法计算格式:I算法公式

对一维C0问题的高次有限元后处理中超收敛计算的EEP(单元能量投影)法提出改进的最佳超收敛计算格式,即用m次单元对足够光滑问题的有限元解答,采用该格式计算的任意一点的位移和应力都可以达到h2m阶的最佳超收敛结果.整个工作分为3个部分,分别给出算法公式、数值算例和数学证明.该文是系列工作的第一部分,针对高次单元提出了凝聚形函数的概念,并证明了相关的逼近定理和等价定理,在此基础上给出了具体的算法公式.     

具有最佳超收敛阶的EEP法计算格式:Ⅲ数学证明

对一维C0问题的高次有限元后处理中超收敛计算的EEP(单元能量投影)法提出改进的最佳超收敛计算格式,即用m次单元对足够光滑问题的有限元解答,采用该格式计算的任一点的位移和应力都可以达到h2m阶的最佳超收敛结果。整个工作分为3个部分,分别给出算法公式、数值算例和数学证明。该文是系列工作的第三部分,对所提出的最佳的EEP超收敛格式给出数学证明。     

具有最佳超收敛阶的EEP法计算格式:Ⅱ数值算例

对一维C0问题的高次有限元后处理中超收敛计算的EEP(单元能量投影)法提出改进的最佳超收敛计算格式,即用m次单元对足够光滑问题的有限元解答,采用该格式计算的任意一点的位移和应力都可以达到h2m阶的最佳超收敛结果。整个工作分为3个部分,分别给出算法公式、数值算例和数学证明。该文是系列工作的第二部分,给出实施算法和数值算例,用以验证理论公式的有效性和正确性。     

Accurate stresses in the thin-layer method

A method is described by means of which accurate strains and stresses can be obtained for problems of wave motion in laminated media modelled with the thin layer method (TLM), a semi-discrete procedure that combines the power of finite elements with that of analytical solutions. It is shown that when the displacements in the TLM are combined with the consistent stresses at the layer interfaces, strains and stresses anywhere in the medium can be obtained with the same level of accuracy as the displacements. The proposed method thus circumvents the intrinsic problem that arises when strains are obtained via differentiation. As a bonus, it also renders the stresses continuous across layer interfaces, which is not the case when stresses are obtained via differentiation of the primary interpolation field. Copyright © 2004 John Wiley & Sons, Ltd.     

A Mixed-Approach Analysis of Deformations in Pipe Bends. Part 2. Three-Dimensional Bending with Internal Pressure

We analyze a problem of determination of stressed state and flexibility of a pipe bend under a combined action of internal pressure and bending moments applied both in and out of plane of the bend curvature. This problem is referred to as a geometrically nonlinear one where an increase in internal pressure brings about a decrease in ovalization stresses induced by bending moments. A unique method of allowing for the pressure is put forward, which is based on the consideration that the pressure-induced circumferential stresses greatly exceed the stresses due to bending moments. A general set of equations for all displacements and stress components is derived. The present results are compared with those reported elsewhere.     

Modified Eshelby tensor for an ellipsoidal inclusion imperfectly embedded in an infinite piezoelectric medium

The modified Eshelby tensor for predicting the effective moduli of particle-reinforced piezoelectric composites is derived for the problem of an ellipsoidal inclusion which is imperfectly bonded to the matrix. A linear interface relation is adopted, which involves discontinuities of the mechanical displacements and electric potential across the interface, and assumes that the corresponding jumps are proportional to the continuous stresses and electric displacements at the interface. The piezoelectric field induced by a uniform eigenstrain given only in the inclusion is deduced analytically. As the induced piezoelectric field is no longer uniform, the average strains and electric displacements are calculated, and the modified piezoelectric Eshelby tensor is evaluated by both an iterative method and a direct method. By comparison, it is shown that the iterative method yields rapidly convergent results.     

Accurate calculation of transverse shear stresses for soft-core sandwich laminates

Accurate evaluation of transverse stresses in soft-core sandwich laminates using the existing 2D finite element (FE) models involves cumbersome post-processing techniques. In this paper, a simple and robust method is proposed for accurate evaluation of through-the-thickness distribution of transverse stresses in soft-core sandwich laminates by using a displacement-based C⁰ continuous 2D FE model derived from refined higher-order shear deformation theory (RHSDT) and a least square error (LSE) method. In this refined higher-order shear deformation theory (RHSDT), the in-plane displacement field for the face sheets and the core is obtained by superposing a global cubically varying displacement field on a zigzag linearly early varying displacement field. The transverse displacement is assumed to have a quadratic variation within the core, and it remains constant in the faces beyond the core. The proposed C⁰ FE model satisfies the condition of transverse shear stress continuity at the layer interfaces and the zero transverse shear stress condition at the top and bottom of the sandwich plate. The nodal field variables are chosen in an efficient manner to circumvent the problem of C¹ continuity requirement of the transverse displacements associated with the RHSDT. The LSE method is applied to the 3D equilibrium equations of the plate problem at the post-processing stage, after in-plane stresses are calculated by using the above FE model based on RHSDT. Thus, the proposed method is quite simple and elegant compared to the usual method of integrating the 3D equilibrium equations at the post-processing stage for the calculation of transverse stresses in a sandwich laminates. The accuracy of the proposed method is demonstrated in the numerical examples through the comparison of the present results with those obtained from different models based on HSDT and 3D elasticity solutions.     

A semi-analytical finite element for laminated composite plates

This paper presents a semi-analytical finite element solution for laminated composite plates. The method is based on a mixed variational principle that involves both displacements and stresses. Finite element meshes are only used in the plane of plate, while the through thickness distributions of displacements and stresses are obtained using the method of state equations. Numerical results show that the rate of convergence of the new method is fast and the solutions can be very close to corresponding exact three-dimensional ones. The use of a recursive formulation of the state equations leads to an algebra equation system, from which solution are sought, whose dimension is independent of the numbers of layers of the plate considered.