Maxwell 3D & Icepak thermal simulation for a air coil

Hi Guys,

I am doing a Maxwell 3D&Icepak thermal simulation for a air coil. I want to get its temperature when loading 1.6A current. So I use Eddy Current type in Maxwell 3D and Temperature and Flow type in Icepak. Then I load 1.6A solid current and adaptive frequency of 800 MHz for the coil. But I get the result as about 75cel. The ambient temperature is 20cel and its temperature rise should be less than 15cel according to its specifications. Do you have any advice for this situation? Thanks!
https://jam8.sapjam.com/embedded_images/show/g6RM6MalHlzsytHHDC9X0B
https://jam8.sapjam.com/embedded_images/show/L3ThZQPUC3poVuCN3SXpHn
https://jam8.sapjam.com/embedded_images/show/tIMGP90aQsfrkCzQNTQgmE

Hi Guys,
I am doing a air coil thermal simulation by Maxwell 3D&Icepak to verify its rated cueernt. First I use Maxwell 3D DC solution type to load DC current and get about 6cel temperature rise in Icepak. But then I want to try this in Maxwell 3D Eddy Current solution type, so I load stranded current with adaptive frequency of 1e-08Hz (which I think is similar to DC current). But I get about 3cel temperature rise in Icepak this time. That is confused for me. I wonder why the temperature rise is different. Could you help raise some advice for this situation? Thanks!

Hi, though I am not expert on this and I have no experience of the software packages you mention, still want to ask some things:
What do you get if you try even lower frequency , something like ##10^{-20}Hz##
What do you get if you try much higher frequency, something like ##1Hz##.

Melon said:

Hi Guys,
I am doing a air coil thermal simulation by Maxwell 3D&Icepak to verify its rated cueernt. First I use Maxwell 3D DC solution type to load DC current and get about 6cel temperature rise in Icepak. But then I want to try this in Maxwell 3D Eddy Current solution type, so I load stranded current with adaptive frequency of 1e-08Hz (which I think is similar to DC current). But I get about 3cel temperature rise in Icepak this time. That is confused for me. I wonder why the temperature rise is different. Could you help raise some advice for this situation? Thanks!

Honestly I wouldn't bother combining EM + Thermal for an air coil. Why? Because the conductive heat transfer of air of so starkly poor.

Also "what frequency DC?" - 1 Hz could be sufficient to be DC; it's "on a spectrum" anyway. For microwave equipment, 50 MHz is "sufficiently DC"! Do a back of the envelope given your calculated inductance of the coil and it's inductive reactance compared to the power source Norton resistance: being within a factor of 10x is already pretty much DC; 100x is easily DC.

The only heat transfer that matters with air at "normal temperatures" is convection and radiation. And very quickly (~300C-500C) radiation dominates. Radiation is far simpler and if it's dominating, it's all you need to worry about.

Convection is highly dependent on the aerodynamics of air flow and odds are a lot of your variability is due to errors in parameters affecting convection which can be chaotic as Navier-Stokes can be chaotic (i.e. laminar vs. turbulent plus dependencies on geometry, gravity, etc.). So it's easy to get variable answers. Also your mesh sizes can create as much error. Always focus on your mesh with FEA.

You can use a hand calculations based on standard pre-defined heat transfer geometries (cubes, cylinders, spheres, slabs, etc.) and get a good answer faster from simply known the IR losses given the coil current levels you are using.

Eddy currents presume and are relevant only for large slabs of coil conductor or core. If you are using wire, there are no eddy currents in the coils and being air there are no eddy currents induced by the coil. If you have really large steel cores that aren't laminated or really large slab coil conductors, sure there will be eddy currents, but even the latter you should be designing the coil/conductors to avoid eddy currents (see how Bitter magnets suppress eddy currents).

The difference between DC solvers like Magnetostatic and Eddy Current is that with Eddy current you need to input the peak value of current. So for example, a Magnetostatic with 1A current input is NOT same as Eddy Current Simulation with 1A current input. For you to have equivalent conditions Eddy current needs to be loaded with 1*sqrt(2) value of current to imitate same condition as in Magnetostatic (or other DC solvers).

If you see the losses here, it will be differing by twice if same value of current is used in these two different type of solvers. This could be leading to different temperature rise.