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 The information provided may be out of date.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.

Apparent inconsistency in induction heating solution methods

Please login with a confirmed email address before reporting spam

Hi everyone,
I'm new to COMSOL 4.3, and I'm trying to properly use the induction heating module. While modeling the problem of a current loop heating several iron inductors embedded in water in 3D, using either External Current Density or Surface Current, I kept getting temperatures that I strongly suspected were too low. I modeled a similar problem of one inductor in axisymmetric 2D, with the intention of comparing heating via Single-Turn Coil Domain with heating from either of the aforementioned methods. I get a result that seems right to me with Single-Turn Coil Domain, but when I deactivate it and activate either External Current Density or Surface Current, I get a temperature rise of less than a degree, a similar wrong result to what I get with 3D modeling.
With External Current Density, I use the current (1000 A) divided by the cross-sectional area of the loop wire, and I use the current divided by the circumference of the loop wire when I use Surface Current instead.
I've attached the 2D axisymmetric version of the test problem that delivers incorrect results. I've slightly modified the materials such that water has a relative permittivity and permeability both of 1, and an electrical conductivity of 0.1 S/m. Can anyone let me know what I'm missing? Thanks in advance!


10 Replies Last Post 2 lug 2014, 17:05 GMT-4

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 18 feb 2013, 08:38 GMT-5
Hi everyone,
Here's a simple 3D version of the problem that has the same issue. Can someone help me with this problem?
Hi everyone, Here's a simple 3D version of the problem that has the same issue. Can someone help me with this problem?


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 mar 2013, 12:19 GMT-5
Hi

I'm not sure it all wrong, but first of all in your 3D model your mesh is too coarse, I get a factor 2.5-3x more induced power by going to "extremely fine" meshing for the small target part, and as your total resistive loss in your coil is only 10 mW I'm not astonished that you get a low induced power either. Now integrating your 13A/mm^2 dissipating over the coil resistance gives about 15W in DC mode, remains to ask what happens at 125 kHz with the impedance ?, could that be the factor roughly 1000 ? Just the section of the coil compared to your target object has a ratio of about (7/0.1) = 4900 so you are not really using efficiently your magnetic field either

by the way you do not seem to use the variable "mu_core", but for me it has a problem of sign for T>320K it becomes negative, not sure that is what you want

note: by using a user defined cylindrical coordinate system it's easier to define the external current along "phi"

Sorry no clear answer I would have to reconstruct it fully, but I do not see why COMSOL should get it wrong, if the model is set-up correctly ;)
--
Good luck
Ivar
Hi I'm not sure it all wrong, but first of all in your 3D model your mesh is too coarse, I get a factor 2.5-3x more induced power by going to "extremely fine" meshing for the small target part, and as your total resistive loss in your coil is only 10 mW I'm not astonished that you get a low induced power either. Now integrating your 13A/mm^2 dissipating over the coil resistance gives about 15W in DC mode, remains to ask what happens at 125 kHz with the impedance ?, could that be the factor roughly 1000 ? Just the section of the coil compared to your target object has a ratio of about (7/0.1) = 4900 so you are not really using efficiently your magnetic field either by the way you do not seem to use the variable "mu_core", but for me it has a problem of sign for T>320K it becomes negative, not sure that is what you want note: by using a user defined cylindrical coordinate system it's easier to define the external current along "phi" Sorry no clear answer I would have to reconstruct it fully, but I do not see why COMSOL should get it wrong, if the model is set-up correctly ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 4 mar 2013, 09:22 GMT-5
Hi Ivar,

Thanks for helping with my problem. The function "mu_core" is a leftover from a previous version of the problem, which involves a relative magnetic permeability that varies with temperature. (I had not considered that it is negative at high temperatures; I'll have to address that in future modeling.) I eliminated that feature so my example file included only the basics.

I had experimented with different meshes for Surface Current, but I found that the final temperature did not converge to a specific value with mesh size; it increased by a uniform amount in each step when I increased the mesh from "normal" up to "extremely fine."

By the way, are you aware of limitations of the induction heating module at low currents or ~100 kHz frequencies? My research group has an industrial induction heater that is capable of turning small steel objects red-hot within seconds, while acting at a frequency of ~100 kHz and less than 20 amperes (corresponding, with the induction coil we are using, to an H-field of approximately 1000 A/m at the current peak). However, I am only able to produce significant heating with COMSOL modeling when the input current is at least an order of magnitude above what the machine is putting out.

Thanks for your help!
Hi Ivar, Thanks for helping with my problem. The function "mu_core" is a leftover from a previous version of the problem, which involves a relative magnetic permeability that varies with temperature. (I had not considered that it is negative at high temperatures; I'll have to address that in future modeling.) I eliminated that feature so my example file included only the basics. I had experimented with different meshes for Surface Current, but I found that the final temperature did not converge to a specific value with mesh size; it increased by a uniform amount in each step when I increased the mesh from "normal" up to "extremely fine." By the way, are you aware of limitations of the induction heating module at low currents or ~100 kHz frequencies? My research group has an industrial induction heater that is capable of turning small steel objects red-hot within seconds, while acting at a frequency of ~100 kHz and less than 20 amperes (corresponding, with the induction coil we are using, to an H-field of approximately 1000 A/m at the current peak). However, I am only able to produce significant heating with COMSOL modeling when the input current is at least an order of magnitude above what the machine is putting out. Thanks for your help!

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 4 mar 2013, 14:35 GMT-5
Hi

well are you sure you have the correct numer of turns in your coil ?,
the only issue I see could be linked to the resonance frequency of the coil and its impedance, that is frequency dependent, check that your model matches the R and L values of your coil

My test with the mesh was to only mesh the target "extremely fine" I didnt change the rest, that made quite some difference, but still not what you are talking about ;)

--
Good luck
Ivar
Hi well are you sure you have the correct numer of turns in your coil ?, the only issue I see could be linked to the resonance frequency of the coil and its impedance, that is frequency dependent, check that your model matches the R and L values of your coil My test with the mesh was to only mesh the target "extremely fine" I didnt change the rest, that made quite some difference, but still not what you are talking about ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 6 mar 2013, 13:44 GMT-5
Hi Ivar,

I tried modeling the loop as a hollow copper tube, instead of like a solid piece like in the original 2D axisymmetric model I uploaded. Changing the thickness of copper in a tube of constant outer diameter does not change the results for heating when Single-Turn Coil Domain is used, so that seems to imply that resistance of the loop does not affect the results. Do you think other parameters that we aren't accounting for could possibly be at work?

The model I referred to earlier (reproducing the induction heater used by our group) is almost exactly the same as the model I uploaded, with the exception that I am using 6 loops and the loop diameter is a little different. I can upload that one if you feel it's relevant; do you think it's better to work out this simple one-loop model first? (We also have a single-loop inductor for our heater that is easily capable of making a nail red-hot at ~15 amperes and ~300 kHz.)

Thanks !
Hi Ivar, I tried modeling the loop as a hollow copper tube, instead of like a solid piece like in the original 2D axisymmetric model I uploaded. Changing the thickness of copper in a tube of constant outer diameter does not change the results for heating when Single-Turn Coil Domain is used, so that seems to imply that resistance of the loop does not affect the results. Do you think other parameters that we aren't accounting for could possibly be at work? The model I referred to earlier (reproducing the induction heater used by our group) is almost exactly the same as the model I uploaded, with the exception that I am using 6 loops and the loop diameter is a little different. I can upload that one if you feel it's relevant; do you think it's better to work out this simple one-loop model first? (We also have a single-loop inductor for our heater that is easily capable of making a nail red-hot at ~15 amperes and ~300 kHz.) Thanks !

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 6 mar 2013, 14:09 GMT-5
Hi

for me the main issue of your original file was the mesh of the target, not the source, surface current or bulk is not canging anything, but the way you calculate the current yes, how many turns, which current density in the coil, wire diameter and overall size. Then the frequency, is it correct, Finally is the surrounding lage enouh to allow the field to loop around ?

Have you tried to compare the results of COMSOL with analytical results ?

--
Good luck
Ivar
Hi for me the main issue of your original file was the mesh of the target, not the source, surface current or bulk is not canging anything, but the way you calculate the current yes, how many turns, which current density in the coil, wire diameter and overall size. Then the frequency, is it correct, Finally is the surrounding lage enouh to allow the field to loop around ? Have you tried to compare the results of COMSOL with analytical results ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 15 mar 2013, 20:33 GMT-4
Hi Ivar,
Sorry for taking so long to reply; I've been busy with deadlines lately. I was able to converge toward a solution by decreasing the mesh size to an extremely small value; however, the heating is much less than what I expect from experiments with our induction heater. I followed your advice of increasing the volume surrounding the heater and workpiece to make room for the magnetic field lines. This increased the maximum temperature, but it is still too low. Also, the result from Single-Turn Coil Domain does not agree with the results from External Current Density and Surface Current. I followed your advice of obtaining analytical results with which to compare, but the analytical results also suggest a much higher maximum temperature than what I have been able to find. (I used the formulas provided here: www.thermopedia.com/content/874/ They seem to agree that what I've found experimentally.)
To sum up, the heating from Single-Turn Coil Domain is significantly lower than what I expect, and the other methods of calculating heating are even lower. I am attaching the model of our induction heater to this post.
Thanks in advance for your help, and thanks again for your time and patience so far!
Hi Ivar, Sorry for taking so long to reply; I've been busy with deadlines lately. I was able to converge toward a solution by decreasing the mesh size to an extremely small value; however, the heating is much less than what I expect from experiments with our induction heater. I followed your advice of increasing the volume surrounding the heater and workpiece to make room for the magnetic field lines. This increased the maximum temperature, but it is still too low. Also, the result from Single-Turn Coil Domain does not agree with the results from External Current Density and Surface Current. I followed your advice of obtaining analytical results with which to compare, but the analytical results also suggest a much higher maximum temperature than what I have been able to find. (I used the formulas provided here: http://www.thermopedia.com/content/874/ They seem to agree that what I've found experimentally.) To sum up, the heating from Single-Turn Coil Domain is significantly lower than what I expect, and the other methods of calculating heating are even lower. I am attaching the model of our induction heater to this post. Thanks in advance for your help, and thanks again for your time and patience so far!


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 18 mar 2013, 10:45 GMT-4
Update:
I also used Coil-Group Domain to attempt to accurately model the heating, but obtained similar results to Single-Turn Coil Domain. Analytical calculations suggest that the power generation should be nearly two orders of magnitude greater than what COMSOL has calculated so far. Any further help is greatly appreciated!
Update: I also used Coil-Group Domain to attempt to accurately model the heating, but obtained similar results to Single-Turn Coil Domain. Analytical calculations suggest that the power generation should be nearly two orders of magnitude greater than what COMSOL has calculated so far. Any further help is greatly appreciated!

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27 giu 2014, 14:14 GMT-4
hi Gregory,

Did you find a solution?

Thank you

Patrice Gouin
hi Gregory, Did you find a solution? Thank you Patrice Gouin

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 lug 2014, 17:05 GMT-4
I did find a solution to the problem; it appears that in 3D mode with either External Current Density or Surface Current, the magnetic field is not calculated correctly and depends on the mesh. My workaround was to either increase the current artificially until the magnetic field in the center of the geometry was as expected or to simply dispense with external current loops altogether and define a uniform magnetic field amplitude in Magnetic Field. The other issue is that the information provided by the induction heater manufacturer was incorrect; the current in the coil is quite a bit higher than 15 A.
I did find a solution to the problem; it appears that in 3D mode with either External Current Density or Surface Current, the magnetic field is not calculated correctly and depends on the mesh. My workaround was to either increase the current artificially until the magnetic field in the center of the geometry was as expected or to simply dispense with external current loops altogether and define a uniform magnetic field amplitude in Magnetic Field. The other issue is that the information provided by the induction heater manufacturer was incorrect; the current in the coil is quite a bit higher than 15 A.

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.