IMPOSING LINEAR CONSTRAINTS IN FINITE-ELEMENT ANALYSIS

被引:17
作者
WEBB, JP
机构
[1] McGill Univ, Montreal, Que
来源
COMMUNICATIONS IN APPLIED NUMERICAL METHODS | 1990年 / 6卷 / 06期
关键词
D O I
10.1002/cnm.1630060607
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
It is often necessary to impose linear constraints on solutions obtained with the finite-element method. This can be done by the penalty method or by eliminating dependent variables. The penalty method requires the careful selection of a penalty parameter and may lead to an inexact imposition of the constraints. Elimination is preferable because it is exact. The zero rows that elimination generates in the coefficient matrix can be removed by reducing the order of the matrix, or by adding back the constraints. The latter approach has the advantage of keeping the imposition of constraints separate from matrix assembly. However, straightforward addition of multi-point constraint equations destroys the symmetry of the matrix. A new method of adding the constraints preserves matrix symmetry.
引用
收藏
页码:471 / 475
页数:5
相关论文
共 7 条
[3]  
FRONCIONI A, 1988, THESIS MCGILL U
[4]  
KANELLOPOULOS VN, 1988, THESIS MCGILL U
[5]   A DIRECT 3-DIMENSIONAL FINITE-ELEMENT METHOD FOR THE SOLUTION OF ELECTROMAGNETIC-FIELDS IN CAVITIES [J].
KONRAD, A .
IEEE TRANSACTIONS ON MAGNETICS, 1985, 21 (06) :2276-2279
[6]  
LOWTHER DA, 1985, COMPUT AIDED DESIGN, P143
[7]  
Zienkiewicz O. C., 1977, FINITE ELEMENT METHO