aomedina - 2006-02-12

Aside from bad meshes, I keep getting into time step problems. Like I mentioned before, there is a general inability to control them. Often times you see the time step grow smaller until it hit your limit of e-80. Then it grows withing a time step or two to a larger number like .01 or larger. However the .pos files get nan output and so you have to trash that and continue playing with the mesh value.  Is there a magical way (some kind of procedure/algorith) that one can use to calculate how large a mesh needs to be? or how refined one can make a mesh for it to reach analysis completion within my lifetime in a cheap student level PC?   I'm asking too much, but maybe you could add suggestions of what to do in these cases.