Discussion Forum

Hi

what about taking a look at the ACDCmodlib.pdf document file. I would start with the example: "Linear Electric Motor of the Moving Coil Type1" or thereaftr, a few of these examples has multipe turn cases, even if they are perhaps one level more complex than what you should go for as a first model, but I expect that soon, you will be building such models too. There are a lot of goodies in these documents.

Good luck
Ivar

Thank you very much for your suggestions, Ivar. Unfortunately the examples in the ACDCmodlib.pdf document file doesn´t show anything can help me. In the "Linear Electric Motor of the Moving Coil Type1" example, coils are not connected. Currents in the coils are known and are imposed in the subdomains settings as Iin and -Iin. On the other hand, in the "Generator in 2D" example, winding sections are not connected in the 2D geometry. The total voltage generated is caculated as the sum of the induced voltages on the individual conductors of the stator, considering alternative polarities (+ or -) from one conductor to the next one.

I´m not sure, but I think I need to connect the coils in my geometry, because the induced current in the loop is what I´m looking for, as this current will enforce or reduce the magnetic field generated by other cables. Furthermore, my second problem is that I need to insert a capacitor in the loop, to compensate its inductive reactance and get higher induced current in the loop. That´s why I´m interested in connecting coils.

Any suggestion will be welcome.

Thanks in advance.

Hello again. I have tried to model 1 turn loop. The geometry is simple, 2 coils at a certain distance. To connect both coils to get the loop I have integred Jz_emqa in both subdomains, obtaining 2 currents I1 and I2. Then, I haved defined a certain voltage V in global equations, such as I1+I2=0. As both coils are in paralell, this voltage V is applied on both coils in the subdomains settings. In this way the loop works ok. So I have something to start with. But with this method I don´t know how to make a 2 turns loop yet.

Please, if anyone could suggest me something I would be grateful.
Thaks in advance.

Hello again. I just want to say that I´m trying to find a solution to my problem, but I don´t find it. If anyone can suggest me something I would be grateful.

Thanks in advance.

Hi

With COMSOL you can really do about everything, but as with most thing you need to have some clear ideas about what you have, what you need, and where you want to go. Your explanations so far does not help me enough to understand really what you are trying to do, hence difficult to exchange and propose something.

Your sketch is a good starting point, I understand its ACDC with currents, but static, quasi static ... ? you talk about capacitive corrections, this gives me the feeling of RF but transient or harmonic needs, are we still in the ACDC domain, and what about skin effects ? (we could start without and in ACDC no?)

can you go with 2D, 2D-axi or must we work in 3D (I strongly suggest to start with 2D). What is constant and what is changing/moving ? Only ACDC coupling or also thermal ? ...
Note that electromagnetics is slightly tricky for your mind, as a current in one plane means magnetic field in the two perpendicular planes, therefore 2D is really a pseudo 3D but with distinguished in-plane or out of plane currents or fields

You talk about two coils, as I understand these are two/multi turns (?) then you can normally simulated them as a single turn by adapting area/volume and current. but if you want to couple between two separate coils you need to simulate them as two separate items

Personally I would start with 2D axisymmetry, Quasi static Magnetics azimutal current (emqa), the best I can guess from your text and sketches, then what next ?
that is coordonates r phi z, dependent variables: Aphi

This means the transform from 2D to 3D is throuigh the (2*pi*r) multiplicant.
This defines the physics (COMSOL shows the formulas nicely for ACDC) and you need either Vloop or Je_phi the current density to define subdomain variables.

To start with a 2-turn or a 2*one-turn is probably the same, in the sens you define Vloop = V_per_turn and/or Je_phy = N*I_wire/Area_wire*Fill_factor (typically Fill_factor is somehere between 55-66%) if you want to go to the "engineering" level immediatly.

Then if you want to move (physically along r,z) one coil w.r.t. the other one need to decide how, parametrised geometry (still lengthy in 3.5a, expecting nice simplfications in 4) with our without CAD link (?) or as ALE or with inequalities over the subdomain ?
How far to move, is it sensitivity analysis or really induction over large displacement ranges ?

Then which parameters to "measure" which output is the goal?
optimising coupling, metrology of motion through ACDC effects ?

Finally one need to introduce the surroundings, as mostly there are other materials around the coils that might introduce side effects, but this we can leave to the end.

My way is to write out as much as I can on paper, to sort/group the items and define an approch in 2-3 steps

Does this help you on the way ?
Hope so, good luck
Ivar

Hello again, Ivar. Thank you very much for your comments. I would first like to apologize for giving insufficient data in my previous messages to describe my problem and also for my English.
What I´m trying to do is to model a magnetic field mitigation system installed in the proximities of underground three-phase power cables in flat formation, which causes the magnetic field I want to mitigate. The mitigation system is done by installing low voltage cables that form a loop. In this loop, the magnetic field I want to mitigate induces currents which create an opposite magnetic field that cancels the field generated by the power cables. To get better mitigation efficiency, the loop impedance can be compensated including a capacitor. This capacitor uses to have so large capacitance values, but can be reduce by using a multi-turn loop, as shown in my sketch (see attached file). So my model has to be an ACDC 2D (non axi) quasi-static magnetic problem with perpendicular induction currents (time harmonic analysis). Of course, skin effects must be included. Since I want to use low voltage cables with cross sections about 120mm2, I don´t think that simulate them as a single turn by adapting area/volume and current is a good choice. Also, later I would like to add thermal coupling, so it would be needed to model the insulation and protection layers that form each cable.

My main goal is to “measure” the currents induced in the loop and the final magnetic field at a certain distance. So the main problem is how to model in comsol such configuration of the loop. In a previous message I commented that I modeled one single loop using “global equations” to force two cables to be parallel. But I don’t know how to do it with more turns and, of course, including the capacitance. I have solved this problem analytically, but there are some approximations that I want to verify with comsol. Moreover, it is difficult to include thermal analysis analytically in this arrangement, so approximations included in the IEC-287 standard must be used, but there are several issues in my model that are not considered in this standard, so finite elements are needed to verify this approximations.

I hope I have explained correctly my problem. Any suggestions would be welcome.
Thank you for your help.

Hi

In principle, the way I understand your case, is (to attempt to avoid a full heavy 3D case) that you can calculate the field of the 3 high power cables (extending infinite in let's say X, probably phased 120° apart) in 2D in the Y-Z plane by a harmonic analysis, then select the height Zp where you would put your secondary loop plane and extrude the result in a X-Y 2D plane (new geometry) where you would have your current running in the plane and the Ay defined by the previous calculation and Ax being the extrusion direction. since basically the high power cables can be englobed in a "long" cylinder, while the secondary loop could be a modelled as circle too, and I would start with one equivalent turn.

Havnt reaaly tried it out, I believe it's possible.
Nevertheless two things puzzles me, while I'm studying the ACDC doc:
1) how to define the phase difference between the HV cables, sine you do not have true harmonic analysis set-up in ACDC
2) how to define the in/out ports you would need to adapt your coil and add the capacitor as an equivalent circuit.

Needs some more thinking, from my side, interesting problem anyhow

Good luck
Ivar

also met that problem when I was trying to model a three-phase distribution cable. The system is, one cable has three insulated conductors, each conductor carries one phase and each phase has 120 degree phase difference. I want to model the field distribution in the cable, probably as a function of phase angle. The fact is, I didn't find a way to define the phase difference in conductors...

Hi,

i just want to hear more about pablo ruiz, have you solve it?

to determine the phase difference try this funktion

JezR=ieff*cos(omega*t-(0*pi/3))
JezS=ieff*cos(omega*t-(2*pi/3))
JezT=ieff*cos(omega*t-(4*pi/3))

or maybe you has find the way (since 8 months ago)

regards