From: Alexander Rabtchevich <alexander.rabtchevich@ia...> - 2007-11-12 18:16:04
I'm trying to solve a collision problem of two bodies. While
investigating I've found several rules which are not obvious to me as
I'm not a specialist in FEA. If they are correct, I think it would be
good to have them in a FAQ.
1. Timestep should be less than a time of sound wave propagation via the
smallest element of the model. To estimate it one can calculate
characteristic lengths of elements as diameter of smallest inscribed circle.
2. A body shouldn't move at one timestep more than half of a contact
element thickness. In the other case some nodes can penetrate it and the
problem will not be solved.
3. Contact elements are required only at one of the contacting bodies.
4. Contact elements thickness should not exceed double size of closest
elements. In the other case nodes from that elements would penetrate
contact elements from the very beginning.
5. A body should not contain any nodes which do not belong to any
element. Such nodes can be produced by GMSH as helpers f.i. when
defining center of a arc.
6. If use GMSH, bodies should be modeled with use of physical volumes.
All the volumes of a body should be engaged in single physical volume.
Sorry to say, but I have a problem with one body which explodes itself
without any interaction with the second body and no forces applied. The
body has no contact elements and (I hope) has no orphan nodes which do
not belong to any element, as it is checked with my external program.
All the elements are of type Solid_Iso_4. The program used is Impact is
0.7.5 under Windows XP.
I've tried to see if the nodes which leave the body belong to single
element - it is no so. Each node belongs at least to two elements. Is
there a way to find out what is wrong with the model or with the program?