On the solution of non-linear diffusion equation

On the solution of non-linear diffusion equation
复制标题

非线性扩散方程的求解

DOI:
--
复制
发表时间:
2014
期刊:
--
影响因子:
--
通讯作者:
R. Kouhia
R. Kouhia
中科院分区:
--
文献类型:
--
作者:
Rakenteiden Mekaniikka;R. Kouhia

文献摘要

参考文献

被引文献

相似文献

扩散方程的求解通常通过对方程的空间椭圆部分进行有限元离散来进行,并且通过某种差分格式(通常是梯形规则(Crank-Nicolson)或Lees的无条件稳定半隐式两步算法)对时间依赖性进行积分。常见的过程是使用皮卡德迭代和梯形规则。然而,在高度非线性问题中,皮卡德迭代的收敛速度慢得难以忍受。一个简单的补救措施是使用一致线性化和牛顿法。对于某一类非线性本构模型,一致的雅可比矩阵是不对称的。本文讨论了雅可比矩阵对称部分的使用和组合牛顿型迭代方案。显示了高度非线性扩散问题的数值结果。还给出了有关时间离散化的注释。
Solution of the diffusion equation is usually performed with the finite element discretization for the spatial elliptic part of the equation and the time dependency is integrated via some difference scheme, often the trapezoidal rule (Crank-Nicolson) or the unconditionally stable semi-implicit two-step algorithm of Lees. A common procedure is to use Picard’s iteration with the trapezoidal rule. However, in highly non-linear problems the convergence of Picard’s iteration is untolerably slow. A simple remedy is to use consistent linearization and Newton’s method. For a certain class of non-linear constitutive models the consistent Jacobian matrix is unsymmetric. This paper discusses the use of the symmetric part of the Jacobian matrix and a combined Newton-type iteration scheme. Numerical results of highly non-linear diffusion problems are shown. Also a note concerning temporal discretization is given.
日本儿童和学生对 TIMSS 科学论文式任务的反应特征 (7) - 9 项任务分析结果的趋势 -
DOI: --
发表时间: 2005
期刊: 日本科学教育学会年会論文集 第29号
影响因子: --
作者:
中山 迅;大場裕子;猿田祐嗣
通讯作者: 猿田祐嗣