Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Is Brinkmann stabilized?
Posted Jul 24, 2010, 6:12 p.m. EDT 1 Reply
Please login with a confirmed email address before reporting spam
I was reading a lovely paper by Khodabkshi on Brinkmann equation where he added a bubble function and got gorgeous convergence. This was back in 2005 and bubbles are at least temporarily missing from 4.0.
So I figured I'd added a stabilization technique myself. That's in the attached .mph. To check, I've create a sequence of meshes each getting finer and then wanted to see that as I changed "alpha" from zero to nonzero then I got faster convergence. The stabilization is one recomended in multiple papers by Stenberg.
Nope. Optimal convergence seems to be alpha=0. :-(
Maybe Brinkmann is already stabilized? Or maybe I don't understand weak contributions......
What I **** think**** they mean is the sum of the integrals over the interior of the mesh elements. Within the interior of each element then I have polynomials so I can differentiate to second order without creating delta functions.
But maybe it is computing the integral over everything, including the mesh boundaries. That would be disappointing.
I have second order derivatives in the stabilization term, I really want those to be only calculated on the interior of each element.
Is this written up anywhere and/or suggestions on possible mistakes?
Regards, John
So I figured I'd added a stabilization technique myself. That's in the attached .mph. To check, I've create a sequence of meshes each getting finer and then wanted to see that as I changed "alpha" from zero to nonzero then I got faster convergence. The stabilization is one recomended in multiple papers by Stenberg.
Nope. Optimal convergence seems to be alpha=0. :-(
Maybe Brinkmann is already stabilized? Or maybe I don't understand weak contributions......
What I **** think**** they mean is the sum of the integrals over the interior of the mesh elements. Within the interior of each element then I have polynomials so I can differentiate to second order without creating delta functions.
But maybe it is computing the integral over everything, including the mesh boundaries. That would be disappointing.
I have second order derivatives in the stabilization term, I really want those to be only calculated on the interior of each element.
Is this written up anywhere and/or suggestions on possible mistakes?
Regards, John
Attachments:
1 Reply Last Post Jul 26, 2010, 2:35 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
Built-in stabilization for the Brinkman Equations interface is included in version 4.0a, using streamline and crosswind diffusion. It was not available in earlier COMSOL versions.
To access the stabilization settings, select Show More Options from the View menu in the Model Builder window (the downward-facing triangle to the left in the Model Builder window's toolbar).
Best regards,
Magnus Ringh
COMSOL
Built-in stabilization for the Brinkman Equations interface is included in version 4.0a, using streamline and crosswind diffusion. It was not available in earlier COMSOL versions.
To access the stabilization settings, select Show More Options from the View menu in the Model Builder window (the downward-facing triangle to the left in the Model Builder window's toolbar).
Best regards,
Magnus Ringh
COMSOL