Learning Forum

[Francoo]

Hi guys,

I am going to do some core loss simulation under transient mode in Maxwell, yet it confuses me what is the physical meaning of the time-domain core loss p(t) obtained from the software. For example, I created a 3D model as in:

[Francoo]

BTW, if the core loss model is set to be the "electrical steel" type, the results match. I doubt there are some bugs with the power ferrite type core loss model in transient mode.

[Francoo]

I get it! It is indeed a bug in Maxwell. When the core loss model is "power ferrite", the method in transient mode is equivalent elliptical loop (EEL) method. In the original version of EEL, coeffienct beta (namely Y in the software) should be 2 to form an elliptical loop shape. Yet, the EEL literature also proposes a general solution when beta is not 2, which is neglected by the ANSYS.
So, if I change the Y here to 2, the results are correct......
Literature: A Dynamic Core Loss Model for Soft Ferromagnetic and Power Ferrite Materials in Transient Finite Element Analysis

[ramadhanmh]

Hi Franco, I am having a similar problem when I compared the calculation and simulation results of core loss. Here you said the calculation and simulation would match if you set the model to be the electrical steel. But I am wondering how you could obtain the coefficients for this type of model (Kh, Kc, Ke) if what you have are only the coefficients of power ferrites (Cm, X, Y).
Thanks in advance.

[Francoo]

What is the material type of your core? The "electrical steel core loss model" (Bertotti's loss separation method) is more suitable for laminated material type like electrical steel, amorphous and nanocrystalline, and the "power ferrite model" (Steinmetz Equation) is more useful for isotropic material like ferrite.
Yet, sometimes "power ferrite model" can also be used for laminated core, as the coefficients can be easily fitted from loss curve in the core datasheet provided by manufacturers. If you want to use electrical steel core loss model for a ferrite core, you need to understand that the kh, kc, ke represent coeffiecients of hysteresis loss, classical eddy current loss and excess loss, and the accurate extraction of these coefficients can be refered by Bertotti's paper (Or google bertotti's loss separation method).
For you convenience, there is also a simple way in ANSYS Maxwell to approximately extract those 'Electrical Steel' coefficients. The basic process is that you input several loss curves and the software will automatically fit the three coefficients for you (yet may lack of physical meaning). You can check ANSYS Maxwell help file for more details. Hope that help you.

[ramadhanmh]

The core material that I am using is ferrite. But as you said, Ansys neglects the beta coefficient for power ferrite if it is set to be other than 2. I tried it myself and I think it is true. So for that reason, I used the electrical steel model even though my core material is actually ferrite.
Since at this point I am not too concerned about the physical meaning, I simply used the Ansys curve fitting feature to extract the electrical steel coefficients as you suggested and the core loss result was close to the calculated one.
Thanks a lot for your help!