Learning Forum

[tpaplham]

I am trying to model a piece of alloy heated by high-frequency induction heating, and there appear to be two ways to model the eddy current losses.

1) eddy effects - this results in the core loss pattern (and thus temperature distribution pattern) that I would expect to see for the geometry of the system, but it appears to be independent of frequency (eddy current loss should be proportional to f^2) and gives me losses that are far less than expected. (TOP IMAGE)

2) core loss - this method is definitely proportional to f^2, but removing the eddy effects as suggested when activating core losses means that I no longer see the expected distribution of core losses. (BOTTOM IMAGE)

I guess my question is, how are the losses being calculated under eddy effects if they're not dependent on frequency? How can I get an accurately frequency-dependent loss calculation that also still incorporates eddy effects to get the expected distribution pattern?

[Navya C]

Hi @tpaplham I suggest you the following
1) Check if the loss data is defined in the material definition or not.
2) Reduce the initial mesh size and make it finer.
3) Increase the region size by few times.
4) Check if you have turned on core loss calculation under excitations.

Regards Navya

[tpaplham]

The core loss data is defined in the material definition, and core loss calculation is enabled for the bottom image. According to ANSYS support material, I should deactivate eddy effects when turning on core losses since eddy current calculation is included. This makes sense from a calculation standpoint, but the distribution doesn't make sense. The distribution of core losses shown in the top image (amplification around the edges of the ribbon) is experimentally verified. There are two questions I am trying to resolve:
1) What is different between the "core loss" model and the "eddy effects" model for excitation that causes the loss (Ohmic loss for eddy effects, core loss for core loss) to have such different distributions? (once again, the eddy effect distribution is experimentally verified to be correct).
2) Why doesn't the eddy effect calculation take frequency into consideration? The equation for Ohmic loss is not directly related to frequency except through frequency effects on J and J*, but losses due to eddy currents should be proportional to frequency squared?

(To address points (2) and (3) I have tried doing this with no effect)

[AndyJP]

>eddy current loss should be proportional to f^2
Maybe you are dealing with the penetration depth, which is somewhat similar to skin depth, and is a root inverse of the frequency. so the effective volume is decreased with f. there may be other factors, thats why numerical simulation is important

[HDLI]

Here are some information regarding eddy current loss and core loss.
1). The core loss includes the eddy current loss. There are two different methods to calculate the losses in FEA, that you are using now.
2). For the eddy current loss of using "set eddy effect", Maxwell calculates induced current and ohmic losses with the eddy current effect based on the conductivity and current distribution. Thus, we could see a closed current loop in this 3D model and there are current and losses at two vertical edges.
3). For the core loss of using "set core loss", Maxwell uses core loss model of materials properties, flux density and dynamic core loss equation to calculate the core loss. It may not present the the closed loop for the eddy current, so with the narrow plane (smaller than coil size), the losses distribution might not be perfect.
4). Because of this particular geometry with narrow steel plane and different calculation methods, the core loss distribution would not be very good to present the eddy current. It should be better if using a wide steel plane.
5). I agreed that current distribution will affect the eddy current losses, especially with skin effect if using a high frequency.
Thanks.

HDLI

[irem taş]

how can I use adaptive mesh for eddy current analyse ?