Kai:
With the fine mesh and slow convergence, do you see any error or warning
messages in the .dat or .msg files? Are there any warnings about negative
eigenvalues or elements distorting so much that they turn inside out?
If you do not have any errors or warnings, then you probably have more
discontinuous iterations with the finer mesh - iterations in which nodes
are opening or closing in the contact interface. Is this true?
At 08:34 PM 4/27/00 -0400, you wrote:
>Hi All,
>
>
>I am wondering if somebody can answer my question. I use Abaqus to
>simulate the contact problem.
>It is simple problem (Berkovich indentation, rigid indenter). I found that
>when the mesh is too fine, the
>simulation is hard to converge. I use UMAT to model my material. The UMAT
>has been verified. When
>the element size becomes large, the simulation converges very quick. Who
>can help me to explain this?
--
----------------------------
Anthony J. Petrella
University of Pittsburgh
[log in to unmask]
----------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|