Energy conservation error in the material point method for solid mechanics

被引:147
作者
Bardenhagen, SG [1 ]
机构
[1] Univ Utah, Dept Mech Engn, Salt Lake City, UT 84112 USA
关键词
stability and convergence of numerical methods; error analysis; basic methods in solid mechanics; linear vibrations;
D O I
10.1006/jcph.2002.7103
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
The material point method (MPM) for solid mechanics conserves mass and momentum by construction, but energy conservation is not explicitly enforced. Material constitutive response and internal energy are carried on discrete points (material points), while the governing equations are solved on an overlying arid. The constitutive response (and internal energy) may be updated at the beginning or end of a numerical time step without affecting mass or momentum conservation properties. Both versions of the algorithm have been applied in the literature. Here energy conservation on the material points is investigated and found to depend strongly on the version of the algorithm used. The energy error is found and partitioned into two terms, one of which is of definite sign (and dissipative). The other term is indefinite, and of opposite sign for the two algorithmic variations. It is shown analytically for a single-material-point free-vibration example that one version of the algorithm is strictly dissipative. For the other version the error terms cancel each other out and energy is conserved. The same trends are borne out in numerical solutions of free axial vibration of continuum bars as the wavelength of the vibrational mode begins to approach the computational cell size. The dissipative algorithm may be described as tending to damp out unresolved modes. For resolved modes, both algorithms give identical results, with no perceptible energy error or dissipation. It is suggested that the dissipative algorithm is a better choice in general, as the damping is consistent with the accuracy of the solution. (C) 2002 Elsevier Science (USA).
引用
收藏
页码:383 / 403
页数:21
相关论文
共 21 条
[1]  
Bardenhagen SG, 1998, AIP CONF PROC, V429, P281
[2]   Dynamic stress bridging in granular material [J].
Bardenhagen, SG ;
Brackbill, JU .
JOURNAL OF APPLIED PHYSICS, 1998, 83 (11) :5732-5740
[3]   The material-point method for granular materials [J].
Bardenhagen, SG ;
Brackbill, JU ;
Sulsky, D .
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING, 2000, 187 (3-4) :529-541
[4]   Numerical study of stress distribution in sheared granular material in two dimensions [J].
Bardenhagen, SG ;
Brackbill, JU ;
Sulsky, D .
PHYSICAL REVIEW E, 2000, 62 (03) :3882-3890
[5]   FLIP - A METHOD FOR ADAPTIVELY ZONED, PARTICLE-IN-CELL CALCULATIONS OF FLUID-FLOWS IN 2 DIMENSIONS [J].
BRACKBILL, JU ;
RUPPEL, HM .
JOURNAL OF COMPUTATIONAL PHYSICS, 1986, 65 (02) :314-343
[6]   FLIP - A LOW-DISSIPATION, PARTICLE-IN-CELL METHOD FOR FLUID-FLOW [J].
BRACKBILL, JU ;
KOTHE, DB ;
RUPPEL, HM .
COMPUTER PHYSICS COMMUNICATIONS, 1988, 48 (01) :25-38
[7]   FLIP MHD - A PARTICLE-IN-CELL METHOD FOR MAGNETOHYDRODYNAMICS [J].
BRACKBILL, JU .
JOURNAL OF COMPUTATIONAL PHYSICS, 1991, 96 (01) :163-192
[8]   THE RINGING INSTABILITY IN PARTICLE-IN-CELL CALCULATIONS OF LOW-SPEED FLOW [J].
BRACKBILL, JU .
JOURNAL OF COMPUTATIONAL PHYSICS, 1988, 75 (02) :469-492
[9]   MASS MATRIX FORMULATION OF THE FLIP PARTICLE-IN-CELL METHOD [J].
BURGESS, D ;
SULSKY, D ;
BRACKBILL, JU .
JOURNAL OF COMPUTATIONAL PHYSICS, 1992, 103 (01) :1-15
[10]  
FUNG Y. C., 1965, Foundations of solid mechanics