Posted:
8 years ago
Dec 19, 2009, 1:50 PM EST

Hi,

I realized that there is an easier way. I calculated the DC resistance from the following formula:

Resistance_DC = length * (resistivity/fill factor) / cross-section area

where the fill factor is the amount of copper in the cross section of the coil domain and the length is calculated as the coils volume divided by its cross-section area. This method gave an reasonable value of the resistance.

//Johan

Hi,
I realized that there is an easier way. I calculated the DC resistance from the following formula:
Resistance_DC = length * (resistivity/fill factor) / cross-section area
where the fill factor is the amount of copper in the cross section of the coil domain and the length is calculated as the coils volume divided by its cross-section area. This method gave an reasonable value of the resistance.
//Johan

Posted:
8 years ago
Dec 19, 2009, 5:24 PM EST

Hi

I agree, if you are in AC/DC static and not interested in higher frequency effects nor in theral effects.

Basically for axis-symmetry, the length of a loop is L=2*pi*r, the current cross section area is a simple integration coupling variable over the subdomain of "1" (implicit int(int(1,dr),dz)), while the volume is an integration couling variable over the subdomain of "2*pi*r (implicit int(int(2*pi*r,dr),dz)). So as you say the average loop length is Volume/Area or simply L_ave= 2*pi*(r_min+r_max)/2.

Now if you try a 2D axi "emqa" (azimuthal current) analysis in static mode you can either apply a loop voltage and get the total current by integrating the results over the coil section, and you should check by hand that you get U=R*I, with I=int(int(Jphi_emqa,r),z) = boundary integration over coil section area of "Jphi_emqa", and R[Ohm]=Lave[m]/(AreaCoil[m^2])/(sigma[s/m]), with sigma the conductivity = 1/(resisitivty[Ohm*m])

or you apply a curret density Je_phy, and get the loop voltage drop through the total energy (power) of the magnetic field (provided you are integration on a large enough area to catch most of the field in the coil and in the air), see the ACDCmodellib doc "Inductance in a Coil" example, or via the resisitve heating over the volume which is Q=U*I=R*I^2

If you apply your good engineer value 55% fill factor for a N multiple winding coil and you get your

N*Area_wire=Area_coil*Fill_Factor.

This gets slightly more complex if you want to consider thermal effects in the coil and or skin effects, but this menas also transient or harmonic analysis, then the coil windings should be modelled in more detail

Pls carefully check when you select/deselect the 2*pi*r multiplier = "Compute volume integral" in the postprocessing subdomain integral window

good luck

Ivar

Hi
I agree, if you are in AC/DC static and not interested in higher frequency effects nor in theral effects.
Basically for axis-symmetry, the length of a loop is L=2*pi*r, the current cross section area is a simple integration coupling variable over the subdomain of "1" (implicit int(int(1,dr),dz)), while the volume is an integration couling variable over the subdomain of "2*pi*r (implicit int(int(2*pi*r,dr),dz)). So as you say the average loop length is Volume/Area or simply L_ave= 2*pi*(r_min+r_max)/2.
Now if you try a 2D axi "emqa" (azimuthal current) analysis in static mode you can either apply a loop voltage and get the total current by integrating the results over the coil section, and you should check by hand that you get U=R*I, with I=int(int(Jphi_emqa,r),z) = boundary integration over coil section area of "Jphi_emqa", and R[Ohm]=Lave[m]/(AreaCoil[m^2])/(sigma[s/m]), with sigma the conductivity = 1/(resisitivty[Ohm*m])
or you apply a curret density Je_phy, and get the loop voltage drop through the total energy (power) of the magnetic field (provided you are integration on a large enough area to catch most of the field in the coil and in the air), see the ACDCmodellib doc "Inductance in a Coil" example, or via the resisitve heating over the volume which is Q=U*I=R*I^2
If you apply your good engineer value 55% fill factor for a N multiple winding coil and you get your
N*Area_wire=Area_coil*Fill_Factor.
This gets slightly more complex if you want to consider thermal effects in the coil and or skin effects, but this menas also transient or harmonic analysis, then the coil windings should be modelled in more detail
Pls carefully check when you select/deselect the 2*pi*r multiplier = "Compute volume integral" in the postprocessing subdomain integral window
good luck
Ivar

Posted:
8 years ago
Feb 1, 2010, 9:57 AM EST

Hi Ivar,

I would like to thank you for your clarifications on DC resistance calculations a while ago...

When it comes to harmonic analysis of multiple winding coils I would like to suggest reading David Meeker’s paper "Continuum Representation of Wound Coils via an Equivalent Foil Approach" (

www.femm.info/examples/prox/notes.pdf) where the coil is represented as a single domain using complex conductivity and permeability. I have not set this kind of model up in COMSOL yet, so I am not sure about how well is works, but it sounds interesting.

//Johan

Hi Ivar,
I would like to thank you for your clarifications on DC resistance calculations a while ago...
When it comes to harmonic analysis of multiple winding coils I would like to suggest reading David Meeker’s paper "Continuum Representation of Wound Coils via an Equivalent Foil Approach" (http://www.femm.info/examples/prox/notes.pdf) where the coil is represented as a single domain using complex conductivity and permeability. I have not set this kind of model up in COMSOL yet, so I am not sure about how well is works, but it sounds interesting.
//Johan

Posted:
8 years ago
Feb 2, 2010, 5:58 AM EST

Hi Johan

that looks promishing, will have a detailed look

Thanks again

Ivar

Hi Johan
that looks promishing, will have a detailed look
Thanks again
Ivar

Posted:
7 years ago
Aug 23, 2010, 6:23 AM EDT

Hi Johan,

now again a while ago: Have you implemented such a model? I am currently facing the same problem, and have to implement it in V4.0a.

Did you achieve you any results that you could share? That would be highly appreciated.

Update to make it clearer:

I'd model the coil windings anyhow as geometry. But I am concerned about the inner structure of the windings - we are going to use litz wires, and I'd have to cope with the remaining skin and proximity effects in the single turns.

Best regards

Matthias

Hi Johan,
now again a while ago: Have you implemented such a model? I am currently facing the same problem, and have to implement it in V4.0a.
Did you achieve you any results that you could share? That would be highly appreciated.
Update to make it clearer:
I'd model the coil windings anyhow as geometry. But I am concerned about the inner structure of the windings - we are going to use litz wires, and I'd have to cope with the remaining skin and proximity effects in the single turns.
Best regards
Matthias