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.

Problem with meshing in a microfluidic channel.

Please login with a confirmed email address before reporting spam

Hey all,

I am trying to simulate the deflection of a half-sphere, with ALE Fluid-structure interaction, stuck on the bottom surface of a microfluidic channel (height = 50 um and width = 300 um) and it works fine with mean inlet flow velocity 10 cm/sec (I see the sphere deflecting), but when I try to increase it to 15 cm/sec, for example, I got an error, stating:

"Non-linear solver didnt converge. Attempt to evaluate real square root of negative number..."

And then a huge mesage follows...

I tried to make the mesh finer, to use adaptive mess, or to use auto-meshing, but every time I had the same error.

So, what do you suggest me to do in order to have higher velocities?


I am also trying to attach the file, but it says it is too big for attaching...

Thank you alot in advance.

3 Replies Last Post 13.03.2013, 10:21 GMT-4
Josh Thomas Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 12.03.2013, 10:46 GMT-4
Dimitrios-

Just an initial thought. How do you know it is necessarily a problem with meshing?

From what you've mentioned, I'd guess the physics (ie the higher velocity) is causing the convergence difficulty. Perhaps this could be corrected by tweaking the mesh, but not necessarily. Maybe you need to tweak something else. Maybe the material model? Maybe the physics set-up? Maybe the solver set-up?

Also, if you want to attach your model, first delete any solutions and meshes. In the file menu click Edit>Clear Meshes and Edit>Clear Solutions. Then, save. You should be able to attach the model in this case.

Best regards,
Josh Thomas
AltaSim Technologies
Dimitrios- Just an initial thought. How do you know it is necessarily a problem with meshing? From what you've mentioned, I'd guess the physics (ie the higher velocity) is causing the convergence difficulty. Perhaps this could be corrected by tweaking the mesh, but not necessarily. Maybe you need to tweak something else. Maybe the material model? Maybe the physics set-up? Maybe the solver set-up? Also, if you want to attach your model, first delete any solutions and meshes. In the file menu click Edit>Clear Meshes and Edit>Clear Solutions. Then, save. You should be able to attach the model in this case. Best regards, Josh Thomas AltaSim Technologies

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 13.03.2013, 04:35 GMT-4
Dear Mr. Thomas,

thank you very much for the help! I noticed for higher Young Modulus values, it works fine for higher velocities, but on the other hand how can I change the physics's or the solver's set up? I am not pretty sure that I understand completely!

I also managed to attach the file, so you can have a better look!

I am looking forward a positive feedback and thank you again beforehand!


Best regards

Dimitrios
Dear Mr. Thomas, thank you very much for the help! I noticed for higher Young Modulus values, it works fine for higher velocities, but on the other hand how can I change the physics's or the solver's set up? I am not pretty sure that I understand completely! I also managed to attach the file, so you can have a better look! I am looking forward a positive feedback and thank you again beforehand! Best regards Dimitrios


Josh Thomas Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 13.03.2013, 10:21 GMT-4
Dimitrios-

My point was that non-convergence can be caused by a number of issues. You shouldn't assume that the problem is with the mesh unless you know for sure. Otherwise you may be missing the real issue.

It makes sense that model changes that result in more deformation of the solid in the physical system are going to give the solver more difficulty (ie lower modulus, higher fluid velocity, etc). The less deformation there is the more you are approaching a simple fluid flow problem which would be easier to solve.

If you haven't gone through the Model Library problems for FSI, I would highly recommend studying those examples. There is some good information on recommended discretization settings based on the Reynolds number of your flow that may apply to your issue. I'm not sure, but regardless the practice problems teach a lot.

I'm not an expert in FSI. Maybe someone else could comment on the specifics of how you might tweak this type of model to obtain convergence for high deformation problems.

FYI - The model you attached converges nicely to a solution with E = 1000[Pa] and vel=40.33[cm/s]. How low did you want to go with the modulus? How high did you want to go with the flow velocity? That information would be helpful if you'd like someone to comment on getting convergence for the more extreme case.

Regards-
Josh
Dimitrios- My point was that non-convergence can be caused by a number of issues. You shouldn't assume that the problem is with the mesh unless you know for sure. Otherwise you may be missing the real issue. It makes sense that model changes that result in more deformation of the solid in the physical system are going to give the solver more difficulty (ie lower modulus, higher fluid velocity, etc). The less deformation there is the more you are approaching a simple fluid flow problem which would be easier to solve. If you haven't gone through the Model Library problems for FSI, I would highly recommend studying those examples. There is some good information on recommended discretization settings based on the Reynolds number of your flow that may apply to your issue. I'm not sure, but regardless the practice problems teach a lot. I'm not an expert in FSI. Maybe someone else could comment on the specifics of how you might tweak this type of model to obtain convergence for high deformation problems. FYI - The model you attached converges nicely to a solution with E = 1000[Pa] and vel=40.33[cm/s]. How low did you want to go with the modulus? How high did you want to go with the flow velocity? That information would be helpful if you'd like someone to comment on getting convergence for the more extreme case. Regards- Josh

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.