袁驷, 袁全. 再谈从矩阵位移法看有限元位移精度的损失与恢复1)[J]. 力学与实践, 2020, 42(6): 689-694. DOI: 10.6052/1000-0879-20-106
引用本文: 袁驷, 袁全. 再谈从矩阵位移法看有限元位移精度的损失与恢复1)[J]. 力学与实践, 2020, 42(6): 689-694. DOI: 10.6052/1000-0879-20-106
YUAN Si, YUAN Quan. REVISITING THE LOSS AND RECOVERY OF DISPLACEMENT ACCURACY IN FEM AS SEEN FROM MATRIX DISPLACEMENT METHOD 1)[J]. MECHANICS IN ENGINEERING, 2020, 42(6): 689-694. DOI: 10.6052/1000-0879-20-106
Citation: YUAN Si, YUAN Quan. REVISITING THE LOSS AND RECOVERY OF DISPLACEMENT ACCURACY IN FEM AS SEEN FROM MATRIX DISPLACEMENT METHOD 1)[J]. MECHANICS IN ENGINEERING, 2020, 42(6): 689-694. DOI: 10.6052/1000-0879-20-106

再谈从矩阵位移法看有限元位移精度的损失与恢复1)

REVISITING THE LOSS AND RECOVERY OF DISPLACEMENT ACCURACY IN FEM AS SEEN FROM MATRIX DISPLACEMENT METHOD 1)

  • 摘要: 本文是文献1的续篇。文献1以一维有限元为例,揭示了其误差主要来自于各个单元的"固端解"。其后,基于这一思想的超收敛计算的单元能量投影(element energy projection,EEP)法得以创立和发展,并有效地用于自适应有限元求解。近期的反思发现,前文的思想精华还有发扬空间:既然单元"固端解''是有限元误差的主要来源,就可以用EEP公式简便地事先求出来,从而可以不经有限元计算而一举得到所需的网格划分。本文简要介绍这一最新方法的思路和机理,并给出初步的数值结果。

     

    Abstract: This paper is a revisit of Ref.1, where it is shown that the errors from one-dimensional finite element (FE) results mostly come from the element fixed-end solutions. Based on this concept, the element energy projection (EEP) method for the super-convergence calculation is developed. Moreover, when the EEP technique is applied to the adaptive FE analysis to estimate and control the errors in FE solutions, the solutions satisfying the user pre-set error tolerances in the maximum norm can be obtained. Recent introspection leads to a realization that the essence in Ref.1 has not been fully exploited: since the element fixed-end solutions are the major source of errors, then it is possible to calculate the errors a priori by using the EEP method, immediately generating a desirable mesh without the need for the FE analysis in advance. This paper gives a brief introduction to this novel idea and some initial numerical results are given to show the validity and effectiveness of the proposed technique.

     

/

返回文章
返回