Thank-you for the references.
I have found the problem that was the wrong way on putting the material
condition in some areas of the mesh.
Now I can use a coupled method without any type of stabilization and it
works properly even with a permeability of order 10e-20!!
The only problem I have is that on the corner of my domain, I get a very
high value of velocity (I mean, bigger than the inlet value, so,
locally, mass conservation isn't accomplished), I have thought that,
perhaps doing a locally refinement, this problem is solved. Am I right?
Any advice will be welcome.
Thanks in advance, best regards
Isa
Wolfgang Bangerth wrote:
I use FEM (finite element method). I want to use a stabilized FE method.
And as I have a change in physical properties (between flow in free
media and flow in porous media), I don't know what type of stabilization
I should use. I have tested GLS (Galerkin least squares), SUPG ... but
they don't seem to work properly with a big change in permeability value
(in free media permeability values infinity whereas in porous media it
values about 10e-11).
You don't say what exactly is going wrong, only that it doesn't work
properly. There's not much we can tell you without a more detailed
description.
In general, if you have a problem where the permeability changes by this
much, the equations have a different character in the two parts of the
domain: it's mostly transport dominated in the free flow part, and
diffusion dominated in the porous part. That makes for a tough problem
with which many formulations/solvers will have problems. I'm fairly sure
there is literature on this sort of thing, though. You may, for example,
want to look at the papers by Todd Arbogast at the University of Texas at
Austin who has dealt with this sort of problem before. I don't know if his
papers have a solution to your particular problem, but they might have
interesting links into the literature.
Another person who has worked on this sort of thing is Beatrice Riviere who
was at the University of Pittsburgh and is now (or will soon be) at Rice
University.
Best
Wolfgang
-------------------------------------------------------------------------
Wolfgang Bangerth email: [EMAIL PROTECTED]
www: http://www.math.tamu.edu/~bangerth/
_______________________________________________
