Discrepancy between Elmer and COMSOL results in the magnetic field calculation of a 3D ectromagnetic pump

[heqiang]

Dear Elmer Discussion Forum members,

I am currently working on a magnetic field calculation of an electromagnetic pump using the Elmer finite element software. The pump consists of 6 coils, which are connected in series in two groups of three for each phase of the three-phase AC power supply.

Initially, I created a 2D axisymmetric model using Elmer, and the results were in good agreement with those obtained using the COMSOL finite element software. Subsequently, I created a 3D model, which covered one-sixth of the entire model (i.e., a 60° sector). When the frequency of the three-phase power supply was set to zero, the results from the Elmer 3D model were in good agreement with those obtained using COMSOL. However, when the frequency was set to 50 Hz, the results obtained using Elmer and COMSOL deviated significantly, particularly with respect to the voltage drop across the two coils connected in series.

I suspect that the calculation of the induced voltage in the coils may be incorrect. I have attached the script file used for the calculation and would appreciate any help or advice regarding the possible causes of the discrepancy between the Elmer and COMSOL results.

I have attached the script file used for the calculation, but due to the file size limit, I had to split the mesh files into two compressed files.

[fjimenez]

Hi,

Very nice case you have here, I am still wrapping my head around the fact you have velocity exported from WhitneyAVSolver, I guess I need to spend more time trying to understand what is happening here hehe. One thing though, you need to add an extra condition to your symmetry boundary condition 1.

Just add

A = Real 0

to Boundary Condition 1 and give it a try. It is particular important for surfaces that deal with JxB. I would also add an iterative solver to MagnetoDynamicsCalcFields, I am not sure what solver would be default to.

I would be very interested in how your case performs against Comsol.

Cheers,

[heqiang]

Hi,

Very nice case you have here, I am still wrapping my head around the fact you have velocity exported from WhitneyAVSolver, I guess I need to spend more time trying to understand what is happening here hehe. One thing though, you need to add an extra condition to your symmetry boundary condition 1.

Just add

A = Real 0

to Boundary Condition 1 and give it a try. It is particular important for surfaces that deal with JxB. I would also add an iterative solver to MagnetoDynamicsCalcFields, I am not sure what solver would be default to.

I would be very interested in how your case performs against Comsol.

Cheers,

Thank you so much for your help and your professional insight! I really appreciate your suggestions about adding an extra condition to the symmetry boundary condition 1 and adding an iterative solver to MagnetoDynamicsCalcFields. I actually discovered the same solutions yesterday and implemented them, and the results are in good agreement with COMSOL.

I was just about to come back here to share my solution with everyone, but I saw your reply. Nevertheless, I want to express my gratitude for your willingness to help and your expertise in the field.

Regarding the fact that I have velocity exported from WhitneyAVSolver, it is related to the calculation work we are doing for an electromagnetic pump. If you are interested, we can discuss it further privately.

Thank you again for your assistance.

[fjimenez]

Nice! Good thing it is working now. I just sent you a DM to continue the conversation about this case.

Cheers,