Induction heating using MagnetoDynamics2DHarmonic solver [SOLVED]

Clearly defined bug reports and their fixes
Post Reply
arved
Posts: 27
Joined: 06 Aug 2020, 15:05
Antispam: Yes

Induction heating using MagnetoDynamics2DHarmonic solver [SOLVED]

Post by arved »

Dear Elmer community,

I did some simulations for verification of axisymmetric induction heating and found that the results of the (obsolete) StatMagSolver agree well with the expected analytical solution, whereas the induced heat computed with the newer MagnetoDynamics2DHarmonic solver together with MagnetoDynamicsCalcFields or BSolver shows big deviations.

The investigated case is a cylinder in a long coil, where I analyze the total heat induced in the central part of the cylinder. The analytical value for the total induced heat is 193.24 W. The curves of the induced heat over the radius of the cylinder are plotted here:
induced-heat.png
induced-heat.png (32.23 KiB) Viewed 4759 times
Furthermore, regarding the mgdyn solver, I made the following observations:
- The imaginary part of the magnetic flux density / field strength is zero in the whole domain
- The real part of the magnetic flux density / field strength is constant in the whole cylinder
- Changing the mesh to quadratic elements (by recombination in gmsh) produces a non-zero imaginary magnetic flux density / field strength, even though it is very low and looks weird
- MagnetoDynamicsCalcFields or BSolver produce the output: SolveSystem: Solution trivially zero!

I am using the solver Version: 8.4 (Rev: Release, Compiled: 2020-08-03) downloaded from the link provided on this website.

Attached you can find the the sif files I used for my simulations, the mesh files (in two different resolutions) are uploaded here: https://cloud9.ikz-berlin.de/s/Piej3YjiLxtGHKP

Best regards
Arved
Attachments
case-mgdyn.sif
(2.83 KiB) Downloaded 332 times
case-statmag.sif
(2.57 KiB) Downloaded 303 times
arved
Posts: 27
Joined: 06 Aug 2020, 15:05
Antispam: Yes

Re: Induction heating using MagnetoDynamics2DHarmonic solver

Post by arved »

Dear Elmer-Team,

did you already have a look at this problem?

Can I do anything to help you with the issue? Do you need more information, another test case, ...?

Best regards
Arved
raback
Site Admin
Posts: 4801
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Induction heating using MagnetoDynamics2DHarmonic solver

Post by raback »

Hi Arved,

I just had a look. It turns out that the new module did not find frequency where the old one had it. I guess when we included more different modules it turned out to be more natural to define it in Simulation section, for example. Now it looks for several places for it, but not the Solver section. Unfortunately there was no complaint so this could go unnoticed.

Attached are some modified .sif files that give pretty much the same results. I now use discontinuous fields for the newer solver as this is one of the things where it adds value. The idea is that the postprocessing is common to 2D/3D even though implementation-wise they are rather different.

Attached is also .m file that makes the comparison.

I modified the code so that it will complain if the frequency is not found. There seems to be good agreement on the results. However, whether the sign of the vector potential could be discussed.

Thanx for nicely made test case!

-Peter
Attachments
case-statmag.sif
Modified sif with old module
(2.85 KiB) Downloaded 334 times
case-mgdyn.sif
Modified sif with new module
(3.12 KiB) Downloaded 357 times
compare.txt
Matlab/octave file (change suffix to .m)
(397 Bytes) Downloaded 311 times
arved
Posts: 27
Joined: 06 Aug 2020, 15:05
Antispam: Yes

Re: Induction heating using MagnetoDynamics2DHarmonic solver

Post by arved »

Hi Peter,

thank you for your support! I also ran the test case and now get the following result:
statmag-vs-mgdyn.png
statmag-vs-mgdyn.png (55.1 KiB) Viewed 4316 times
For anyone who is interested in the theoretical background and the analytical solution: Have a look at Chapter 6 in S. Lupi: Fundamentals of Electroheat. Springer, 2017.

Best regards
Arved
Post Reply