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.

3D eddy current, external current density source for non round coil geometries (TEAM21)

Please login with a confirmed email address before reporting spam

Hello Comsol Users,

I work with the AC DC Modul (electric and induction current) and simulate an eddy current Sensor 3D. (for Impedanz...)

To verify the result i want to create a model from "Testing Electromagnetic Analysis Methodes (TEAM 21)

h*ttp://www.compumag.co.uk/team.html (her you can find the problems)

but the coils are not round and i cannot set the external current density source in the subdomain area with Jx = J_0*y/sqrt(x^2+y^2), Jy=...

Futhermore i must set the conductivity of the coil to zero, that i have no eddy current (skin effect) within the coil.

I try it with Port Boundary but then i need a conductivity and that make a skin effect in the coil. This have an effect on the results.

And if i solve also with the variable V (AxAyAz10...psi) i have also much problems with the solver (geometric multigrid)

Have someone an idea to set the external current for non round coils?? Can help the tangential variables t1x t2x... ?

THANKS A LOT!

best regards

Matthias


4 Replies Last Post Nov 25, 2010, 10:32 a.m. EST

Please login with a confirmed email address before reporting spam

Posted: 2 decades ago Jul 14, 2009, 11:25 a.m. EDT
Hi,

Allow me to give some partial answers to the questions you raise:

1) To set the current density in a non-circular coil, I resorted to an
approach in which I specified the spatial distribution of the
current density vector. This approach has limited applicability,
and possibly alternatives easier to apply do exist.

2) To model a winding (also called stranded conductor) you indeed
need to set the conductivity equal to zero. Different people have
succesfully applied this in Comsol Multiphysics and I could send
you example models if of interest to you.
The issue to pay attention to in solving a 3D transient model with
zero-conductivity is that one has to enforce the cauging
condition at each time step in over to enforce the magnetic
flux to remain divergence free. This can be done by solving
simultaneously for vector and scalar potential.

3) I suggest to solve the model using a direct solver first, and worry
about an iterative solution technique in a second stage.

I share your interest in eddy-current computations.

Bye, Domenico Lahaye.
Hi, Allow me to give some partial answers to the questions you raise: 1) To set the current density in a non-circular coil, I resorted to an approach in which I specified the spatial distribution of the current density vector. This approach has limited applicability, and possibly alternatives easier to apply do exist. 2) To model a winding (also called stranded conductor) you indeed need to set the conductivity equal to zero. Different people have succesfully applied this in Comsol Multiphysics and I could send you example models if of interest to you. The issue to pay attention to in solving a 3D transient model with zero-conductivity is that one has to enforce the cauging condition at each time step in over to enforce the magnetic flux to remain divergence free. This can be done by solving simultaneously for vector and scalar potential. 3) I suggest to solve the model using a direct solver first, and worry about an iterative solution technique in a second stage. I share your interest in eddy-current computations. Bye, Domenico Lahaye.

Please login with a confirmed email address before reporting spam

Posted: 2 decades ago Jul 15, 2009, 10:45 a.m. EDT
Hello Domenico Lahaye,

Thanks for your answer and help!

Yes you have right, i think also that the spatial distribution of the current density vector in the subdomain area are limited for symmetries geometries. For 2 years i work with the Ansoft Maxwell Software. In this Software was it easy to specify the current density for a non-circular coil. I only set a current density terminal + Current source and i can set the skin effect off with one click but comsol makes it not so easy to me. I try many to specified the current density but with zero conductivity i see no alternatives to apply the current source.

2) Im very interest for an example with stranded conductor, then I can find out whether my model is correct and can study more with stranded conductor models. Thanks for your offer! It is possible to create a 3D "transient" model with a frequency of 1 MHz?
I know the gauge fixing problem in 3D and read many article and books about it, but I'm not sure whether I've understood the theory correctly.
Mean you with simultaneously solving the variables "psi" and "tAxAyAz10..."?

3)Yes you have right, the iterative solver are not easy to use but it could help to save much memory.
My model have often more as 100000 DOF, because i can exploit in the geometry only one symmetry.
Furthermore i make experimental with infinity elements for a better accurate solution

I'm fascinated by Comsol Multiphysics!

Bye, Matthias





















Hello Domenico Lahaye, Thanks for your answer and help! Yes you have right, i think also that the spatial distribution of the current density vector in the subdomain area are limited for symmetries geometries. For 2 years i work with the Ansoft Maxwell Software. In this Software was it easy to specify the current density for a non-circular coil. I only set a current density terminal + Current source and i can set the skin effect off with one click but comsol makes it not so easy to me. I try many to specified the current density but with zero conductivity i see no alternatives to apply the current source. 2) Im very interest for an example with stranded conductor, then I can find out whether my model is correct and can study more with stranded conductor models. Thanks for your offer! It is possible to create a 3D "transient" model with a frequency of 1 MHz? I know the gauge fixing problem in 3D and read many article and books about it, but I'm not sure whether I've understood the theory correctly. Mean you with simultaneously solving the variables "psi" and "tAxAyAz10..."? 3)Yes you have right, the iterative solver are not easy to use but it could help to save much memory. My model have often more as 100000 DOF, because i can exploit in the geometry only one symmetry. Furthermore i make experimental with infinity elements for a better accurate solution I'm fascinated by Comsol Multiphysics! Bye, Matthias

Please login with a confirmed email address before reporting spam

Posted: 2 decades ago Jul 15, 2009, 12:20 p.m. EDT
Hola Matthias,

There is number of issues to reply you with:

1) a) Setting the spatial distribution of the current density can be done for the coils
of the TEAM problem you are interested in: indeed, J is essentially plus or
minus one on the straight side and equal to the circular coil J in the bend
corners.
b) Possibly one can copy the Ansys trick in Comsol by solving a PDE for the
components of J in a seperate application mode (did not yet understand the
details myself though, support?).
c) The modeling of voltage driven stranded conductors and current driven
solid conductors is discussed in Chapter 2 of my thesis (available from my
home page). Essentially you supply the PDE with an ODE modeling the
electrical excitation (mail me for details).

2) a) I am working on notes explaining one and other. Attached you will find
a pre-release of these notes, including some example in 2D for time being.

b) I do not understand your 3D transient 1Mhz question. The AC/DC module
comes with an example modeling skin-effect you might want to look into.

c) By gauge fixing I do indeed mean what you understood.

Do you still plan to make your hands durty with that TEAM problem?

Bye, Domenico.
Hola Matthias, There is number of issues to reply you with: 1) a) Setting the spatial distribution of the current density can be done for the coils of the TEAM problem you are interested in: indeed, J is essentially plus or minus one on the straight side and equal to the circular coil J in the bend corners. b) Possibly one can copy the Ansys trick in Comsol by solving a PDE for the components of J in a seperate application mode (did not yet understand the details myself though, support?). c) The modeling of voltage driven stranded conductors and current driven solid conductors is discussed in Chapter 2 of my thesis (available from my home page). Essentially you supply the PDE with an ODE modeling the electrical excitation (mail me for details). 2) a) I am working on notes explaining one and other. Attached you will find a pre-release of these notes, including some example in 2D for time being. b) I do not understand your 3D transient 1Mhz question. The AC/DC module comes with an example modeling skin-effect you might want to look into. c) By gauge fixing I do indeed mean what you understood. Do you still plan to make your hands durty with that TEAM problem? Bye, Domenico.


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 25, 2010, 10:32 a.m. EST
Hi Domenico,

I'm having the same problem as Matthias, however with a different coil. I'd like you to send me some examples about the stranded coil solved in multiphisics. I'm trying to solve first the electrical circuit and then the magnetic problem, but I don't know what I am doing wrong.

Thanks,
Iker.
Hi Domenico, I'm having the same problem as Matthias, however with a different coil. I'd like you to send me some examples about the stranded coil solved in multiphisics. I'm trying to solve first the electrical circuit and then the magnetic problem, but I don't know what I am doing wrong. Thanks, Iker.

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.