Discussion Forum

HI, Marouane,

It sounds that you are forgetting the gap in the coils (unless you want a closed-loop condition). Sorry if I am mistaken. Notice that in 2D-axisymmetric, when a simple circle (torus in 3D) is set as a coil domain, COMSOL is considering it having an infinitesimal gap, in order to feed it (with some non-null voltage), but so small that the magnetic field it creates is equal to that of a perfect torus. If you model the coil in 3D, you must include the gap in the geometry, as a small discontinuity (full of air, generally). In the ends of the torus (borders of the gap) you can set both terminal conditions for the coil.

I hope this helps you.

Jesus.

Thanks for your reply Jesus.

But if I understood correctly what you're talking about, I think I've already included the gap. In fact, in the definitions node I added an "Infinite Element domain 1" which is the air around the coils and an "Adjacent 1" which is the coils with "Exterior boudaries" checked in "output entity". Is that what you're talking about? If not, could you be more specific regarding the means to do so please?

Thanks again!

Hi, Marouane,

No, I wasn't talking about the infinite elements around all the geometry (including the surrounding air). I referred to a little breaking (a physical discontinuity) in the conductor itself, unless you are modelling that coil as a secondary closed circuit (I think this is not the case).

Have a look to the documentation of AC/DC module, particularly to "AC/DC Module > Modeling with the AC/DC Module > Coil Domains > Using Coils in 3D Models", or see the "inductor_3d" model.

Thus, in your 3D model, you can do something similar: cut the torus of the primary circuit with two closed planes (for instance, adding, by duplicating the original torus, another torus with only 1 deg for the revolution angle), then extract that gap domain from the study (from the Magnetic Fields physics), and finally setting the gap boundaries with suitable conditions (current / voltage / circuit ...).

I hope this helps you.

Jesus.

Hi

If I understood well the explanation in v4.3a, you do not even need a "gap" in 3D (which is not physical and introduces other artefacts) but you need to have a cut surface to define the Coil total area, and a loop line for the current flow direction

--
Good luck
Ivar

Thanks for the reply.

So, I tried what you said and this is what I did : I created another torus with 1 degree and set it as an air material. Then I excluded it from the mf study. Afterwards, in the "single-turn coil domain" node, I added a "boundary feed" sub-node and set it as the boundary of the new torus, with 1 A coil excitation, and I added a "Ground" sub-node set as the other boundary of the gap torus.

Unfortunatly, I still get the same error when I run the time dependent simulation. I'm getting really desperate :(

Have I done something wrong? Can you help me further pretty please? :(

Hi

first of all you need a block or a sphere to enclose fully your coil to get the field to loop around, not too big to avoid RAM crash, and not too small to allow the field to loop unafected, but MF should cover air + coil !

Then if you have only an average current, or an oscillating current, the easiest torus type coil is obtained by defining a cylindrical coordinates with axis along the torus ring then to add a current density J0 along phi from this coordinate.

The single turn coil is also possible ,bit it reauires more exercicing, too long to explain here, there are some new model library cases.

The multi-turn coil is the most complex and require quite some additional sub-nodes, mostly I need also to add an A gauge. you need to study the help file, then build very simple models and go step ba step, ne physics at the time until you are fully comfortable with all domain and BC settings

--
Good luck
Ivar

Thanks for your reply.

So, I tried what you said to no avail, knowing that I'm already using an air sphere of 1 meter radius around the coil. This is what I did :

- I defined a cylindrical coordinate system then I included in the coil domain a " Single Turn Domain " node and "external current density" node with 1 A/m² along phi. Still, when running the time dependent I get the "Failed to find initial solution" error...

- Like I mentioned above, I used "Single Turn domain " subnodes, "Boundary Feed" and "Ground" , on a gap in the coil. I got also "Failed to find initial solution".

- I followed the "helmholtz-coil" model and I used the "Multi coil domain" node : I selected "Circular" coil type with 1 as a number of turns and I added a " Reference Edge" node which is the inner circular edge of the coil. Guess what? Yes, "Failed to find initial solution"...

I'm starting to think it's some kind of a bug in Comsol which occurs with circular coils in 3D's time dependent simulation. Because I tested a linear vertical coil using one of these BCs and I got the simulation running perfectly...

But I really hope it's not a bug. I hope it's me missing something. And I'm desperately hoping for your experienced help. Please, tell me what's wrong ! :(

Hi

I do not beleive there is a bug (except if you are in the old 4.0 or 4.1, there maybe ;)
There must be something else you have missed. some BC missing such that there is several solutions and the solver cannot decide which to propose, that is the most common error.

If you impose en external current you should NOT add ontop the single or multipl turn coil BCs, that is doubling up

--
Good luck
Ivar

Dear Ivar
I have seen your support in many messages. Hope you can help me.
I am getting error like this any comment please :

Failed to find a solution.
Singular matrix.

There are 3591 degrees of freedom that do not occur in any equation (empty columns in matrix) for the variable comp1.siec.c_pce1_s1.
at coordinates: (0,0,0.00138), (0.00169672,0,0.00138), (0,0,0.00143), (0,0,0.00142167), (0,0,0.00141333), ...
Returned solution is not converged.

s1 is depositing species