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 The information provided may be out of date.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.

Best solver for non-linear, time-dependent superconductor model

Please login with a confirmed email address before reporting spam

Until now, I have used Comsol 3.5a for modelling highly non-linear, time-dependent superconductor problems. The solver I used for this was UMFPACK, which performed best out of all the available options in that version of Comsol. I have now upgraded to Comsol 4.3 and this solver seems to have disappeared.

Does anyone have any advice as to the best solver for this type of problem? And what might be the most suitable substitute for the UMFPACK solver that worked so well previously? I am using the General Form PDE in the Multiphysics module to solve the model, for which the superconductor's electromagnetic properties are modelled using a highly non-linear E-J power law.

3 Replies Last Post 12 août 2012, 07:20 UTC−4

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 9 août 2012, 07:06 UTC−4
Hi Mark,

I use the general form PDE for superconductors modelling as wel. I do not know too much about the solvers, but I test them at times, compare and see what works best. in my case, almost at all times the default solver works. that is - direct, advanced, full coupled, in this order.

Have you managed to couple the PDE with any other physics? I'm trying to couple PDE with MF and I get stuck.

I'm using 4.2a.

best of luck,
Shoval.
Hi Mark, I use the general form PDE for superconductors modelling as wel. I do not know too much about the solvers, but I test them at times, compare and see what works best. in my case, almost at all times the default solver works. that is - direct, advanced, full coupled, in this order. Have you managed to couple the PDE with any other physics? I'm trying to couple PDE with MF and I get stuck. I'm using 4.2a. best of luck, Shoval.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 9 août 2012, 07:40 UTC−4
Hi Shoval,

I've found MUMPS works, but PARDISO and SPOOLES don't. Awaiting a response from Comsol support as to whether there may be a better solver hidden away somewhere.

In 3.5a, I coupled the superconductor's electromagnetic PDE with a PDE for thermal modelling successfully, but haven't attempted this yet using 4.3. I'm currently trying to get a 2D axisymmetric coil model working using curl elements (instead of Lagrange elements), but not having any success.
Hi Shoval, I've found MUMPS works, but PARDISO and SPOOLES don't. Awaiting a response from Comsol support as to whether there may be a better solver hidden away somewhere. In 3.5a, I coupled the superconductor's electromagnetic PDE with a PDE for thermal modelling successfully, but haven't attempted this yet using 4.3. I'm currently trying to get a 2D axisymmetric coil model working using curl elements (instead of Lagrange elements), but not having any success.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 12 août 2012, 07:20 UTC−4
Attached a model that might help you - it helped me.

about the solvers - try looking at the file COMSOLMultiphysicsReferenceGuide.pdf, it should be hidden somewhere in the doc library in your comsol directory. it has much info about the solvers - I plan on reading that soon too.
Attached a model that might help you - it helped me. about the solvers - try looking at the file COMSOLMultiphysicsReferenceGuide.pdf, it should be hidden somewhere in the doc library in your comsol directory. it has much info about the solvers - I plan on reading that soon 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.