Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

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.

Adaptive mesh refinement within a parametric sweep

Please login with a confirmed email address before reporting spam

I am trying to solve a fluid flow problem over a range of inlet flows using a parametric sweep. The solution does not converge at higher flows, so what I would like to do is run the sweep until it fails to converge, then perform an adaptive mesh refinement, and continue iteratively repeating the sweep and mesh refinement until the sweep succeeds. I have not found a way to do this, so any help would be appreciated.

An alternative approach I tried to set up was to run an adaptive mesh refinement following each step of the parametric sweep, but no matter how I set it up, it always runs the parametric sweep inside the adaptive mesh refinement (thus always failing).

Anyone succeed in doing this, or can anyone suggest an alternative approach? I don't know exactly what part of the geometry is causing the problem, so I can't use a locally fine mesh. I tried a very fine mesh over my entire geometry which got me further along the sweep, but took 9 days to solve.

--
Steven Conrad, MD PhD
LSU Health

3 Replies Last Post 27 mar 2013, 16:00 GMT-4
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 mar 2013, 05:20 GMT-4
Hi

You can access the mesh commands under the solver nodes (try a right click), see also below

--
Good luck
Ivar
Hi You can access the mesh commands under the solver nodes (try a right click), see also below -- Good luck Ivar


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 mar 2013, 17:19 GMT-4
Thanks, Ivar.

I looked at the attached images, but they refer to deformed meshes or to a time dependent solver. My problem is different, and will try to explain it a different way.

My underlying problem is that I cannot get convergence during a parametric sweep of inlet flows on a stationary 3D fluid flow problem. I am looking at flows of 100:100:1200 mL/min, and it only converges up to about 500. The Reynold's number is only 1900 at the max flow, but the geometry results in some disturbed flow patterns leading to non-convergence. I suspect I can get convergence with a finer mesh, but I cannot determine what part of the mesh is giving me a problem and needs refinement. So what I would like to do is run the parametric sweep with increasing flow until it fails convergence, then adaptively refine the mesh. Then I would like to restart the parametric sweep. If the sweep succeeds, then I am done. If not, I would like to refine the mesh again and repeat this process until the sweep succeeds.

With this approach, the solver stops when it fails convergence and therefore cannot continue to the adaptive mesh refinement. Is there any way to make this work?

As an alternative, and perhaps a better, approach I would like to run the adaptive solver at each step of the parametric sweep. That is, the sweep would start at 100, solve it, refine the mesh, then move on to 200, refine again, etc. But I could never set up the adaptive mesh refinement to operate inside of the sweep. No matter how I set it up, it always runs the sweep inside the adaptive solver.

--
Steven Conrad, MD PhD
LSU Health
Thanks, Ivar. I looked at the attached images, but they refer to deformed meshes or to a time dependent solver. My problem is different, and will try to explain it a different way. My underlying problem is that I cannot get convergence during a parametric sweep of inlet flows on a stationary 3D fluid flow problem. I am looking at flows of 100:100:1200 mL/min, and it only converges up to about 500. The Reynold's number is only 1900 at the max flow, but the geometry results in some disturbed flow patterns leading to non-convergence. I suspect I can get convergence with a finer mesh, but I cannot determine what part of the mesh is giving me a problem and needs refinement. So what I would like to do is run the parametric sweep with increasing flow until it fails convergence, then adaptively refine the mesh. Then I would like to restart the parametric sweep. If the sweep succeeds, then I am done. If not, I would like to refine the mesh again and repeat this process until the sweep succeeds. With this approach, the solver stops when it fails convergence and therefore cannot continue to the adaptive mesh refinement. Is there any way to make this work? As an alternative, and perhaps a better, approach I would like to run the adaptive solver at each step of the parametric sweep. That is, the sweep would start at 100, solve it, refine the mesh, then move on to 200, refine again, etc. But I could never set up the adaptive mesh refinement to operate inside of the sweep. No matter how I set it up, it always runs the sweep inside the adaptive solver. -- Steven Conrad, MD PhD LSU Health

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27 mar 2013, 16:00 GMT-4
Hi

Sorry, but here I cannot really help that much (try support) I'm fighting with something similar these days too, and need more time to be comfortable with the "dynamic" meshing and re-meshing options, these are features I haven't had to use too often, and things has changed over the recent v4 "evolution" (far more detailed options)

--
Good luck
Ivar
Hi Sorry, but here I cannot really help that much (try support) I'm fighting with something similar these days too, and need more time to be comfortable with the "dynamic" meshing and re-meshing options, these are features I haven't had to use too often, and things has changed over the recent v4 "evolution" (far more detailed options) -- Good luck Ivar

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.