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 The information provided may be out of date.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.

"Nonlinear solver did not converge"

Please login with a confirmed email address before reporting spam

Hi

I am trying to model a tapering tube through which a cell moves along with the liquid. I want to study the fluid as well as the cell dynamics. I used Fluid-Structure interaction, time-dependent solver for the same. It is a fine, physics-controlled mesh. When I run the simulation, there is an error in a short while saying "Nonlinear solver did not converge." And it also says trying to find real sqrt of negative number.
When I try at finer or extremely fine mesh resolution, it says, requested mesh quality could not be achieved.
When the same simulation was carried out in an even tube, without tapering, these errors did not arise.
What could be causing the problem? How do I correct it?

Best
Renu

7 Replies Last Post 28 feb 2012, 18:43 GMT-5
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 2 feb 2012, 01:24 GMT-5
Hi

when you have a "particle" moving in a fluid you will ahve large deformation of the mesh from the motion of the particles, this is probably what is giving you trouble, as you need a fine mesh to resolve the flow aspects, and a coarse mesh to allow it todistort without inversing. These two conditions are obviously in contradiction.

I would suggest that you try first to use the new particle flow, based on the fluid solving alone to give the velocity field and then track how a particle will follol this flow. This assumes in practivce that your fluid is not influenced by the particles and their size, even if this is probably not fully true, its a good starting point. And this would allow you to estimates velocities and mesh size time to defome, and help you to define a "remishing" conditions that is probably the only way out, but it's rather slow to solve ;)

--
Good luck
Ivar
Hi when you have a "particle" moving in a fluid you will ahve large deformation of the mesh from the motion of the particles, this is probably what is giving you trouble, as you need a fine mesh to resolve the flow aspects, and a coarse mesh to allow it todistort without inversing. These two conditions are obviously in contradiction. I would suggest that you try first to use the new particle flow, based on the fluid solving alone to give the velocity field and then track how a particle will follol this flow. This assumes in practivce that your fluid is not influenced by the particles and their size, even if this is probably not fully true, its a good starting point. And this would allow you to estimates velocities and mesh size time to defome, and help you to define a "remishing" conditions that is probably the only way out, but it's rather slow to solve ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 feb 2012, 17:52 GMT-5
Thanks for the reply, Ivar.
I am now trying to model the same with a stationary cell at different points, using laminar flow alone.
The new error is "Expression could not be evaluated. And I found this in the log.
"Time-Dependent Solver 1 in Solver 1 started at 2-Feb-2012 14:48:45.
Nonsymmetric matrix found."
I am not using any variables in the model. The inlet velocity is also constant.
What is going wrong?
Thanks for the reply, Ivar. I am now trying to model the same with a stationary cell at different points, using laminar flow alone. The new error is "Expression could not be evaluated. And I found this in the log. "Time-Dependent Solver 1 in Solver 1 started at 2-Feb-2012 14:48:45. Nonsymmetric matrix found." I am not using any variables in the model. The inlet velocity is also constant. What is going wrong?

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 feb 2012, 01:40 GMT-5

Thanks for the reply, Ivar.
I am now trying to model the same with a stationary cell at different points, using laminar flow alone.
The new error is "Expression could not be evaluated. And I found this in the log.
"Time-Dependent Solver 1 in Solver 1 started at 2-Feb-2012 14:48:45.
Nonsymmetric matrix found."
I am not using any variables in the model. The inlet velocity is also constant.
What is going wrong?


can you please attach the model file. i will try to have a look at it ?
[QUOTE] Thanks for the reply, Ivar. I am now trying to model the same with a stationary cell at different points, using laminar flow alone. The new error is "Expression could not be evaluated. And I found this in the log. "Time-Dependent Solver 1 in Solver 1 started at 2-Feb-2012 14:48:45. Nonsymmetric matrix found." I am not using any variables in the model. The inlet velocity is also constant. What is going wrong? [/QUOTE] can you please attach the model file. i will try to have a look at it ?

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 3 feb 2012, 02:35 GMT-5
Hi

I have noticed sometimes when you have done many changes and tests, COMSOL does no longer update the solver settings accordingly, this happens particularly when you change many things in the solver sequence, which COMSOL does not understand, and decides that you want to control the solver in detail, so it does no longer update everything.

In these cases, my way out is to delete the solver sequence and restart a fresh new one (but take care to save your results tree by pointing them to anot data set)

--
Good luck
Ivar
Hi I have noticed sometimes when you have done many changes and tests, COMSOL does no longer update the solver settings accordingly, this happens particularly when you change many things in the solver sequence, which COMSOL does not understand, and decides that you want to control the solver in detail, so it does no longer update everything. In these cases, my way out is to delete the solver sequence and restart a fresh new one (but take care to save your results tree by pointing them to anot data set) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 6 feb 2012, 18:19 GMT-5
I am not able to upload the file on to this discussion forum.( File extension error)
Thanks Ivar, it helped. I modeled the same with stationary cell and observed the fluid flow, through laminar flow. Now I am trying to give the cell elastic nature through FSI module, and the time step is not converging again. I read in the forums that i need to use saved initial values from stationary solver. But I am not able to store the results from stationary solver and reuse it. How do I do this?
I am not able to upload the file on to this discussion forum.( File extension error) Thanks Ivar, it helped. I modeled the same with stationary cell and observed the fluid flow, through laminar flow. Now I am trying to give the cell elastic nature through FSI module, and the time step is not converging again. I read in the forums that i need to use saved initial values from stationary solver. But I am not able to store the results from stationary solver and reuse it. How do I do this?

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 6 feb 2012, 18:31 GMT-5

I am not able to upload the file on to this discussion forum.( File extension error)
Thanks Ivar, it helped. I modeled the same with stationary cell and observed the fluid flow, through laminar flow. Now I am trying to give the cell elastic nature through FSI module, and the time step is not converging again. I read in the forums that i need to use saved initial values from stationary solver. But I am not able to store the results from stationary solver and reuse it. How do I do this?


Here are my files. "Exp2" is the one with moving cell and "elasticell" is the one with fsi.
www.mediafire.com/?agfull3u2r5f6sb,7dcj9wz0k9tvqeg
[QUOTE] I am not able to upload the file on to this discussion forum.( File extension error) Thanks Ivar, it helped. I modeled the same with stationary cell and observed the fluid flow, through laminar flow. Now I am trying to give the cell elastic nature through FSI module, and the time step is not converging again. I read in the forums that i need to use saved initial values from stationary solver. But I am not able to store the results from stationary solver and reuse it. How do I do this? [/QUOTE] Here are my files. "Exp2" is the one with moving cell and "elasticell" is the one with fsi. http://www.mediafire.com/?agfull3u2r5f6sb,7dcj9wz0k9tvqeg

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28 feb 2012, 18:43 GMT-5
I noticed a similar problem with the FSI, particularly with large deformations. It seems to have been exacerbated in 4.2a from 4.2. I have a feeling that has to do with the tolerance/iteration termination settings but have yet to resolve this issue.

I'd be interested in any possible solutions too.
I noticed a similar problem with the FSI, particularly with large deformations. It seems to have been exacerbated in 4.2a from 4.2. I have a feeling that has to do with the tolerance/iteration termination settings but have yet to resolve this issue. I'd be interested in any possible solutions too.

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.