Comparison of losses computation with Flux and custom solution on Matlab

Ronan CORIN

Hello,

 

I'm currently studying the losses of a ring core, with isotropic Fe-Si material, in Flux 2D.

I wanted to compare the results given by two ways :

• with the loss computation module included in Flux.
• with a custom computation with Matlab or Octave (after extracting flux density values in face elements in Flux)

I'm worrying about my computation method, because they don't give the same results :

 

Global flux density	0.36T	0.90T
Losses : Flux	0.650 W	3.038 W
Losses : Matlab/Octave	0.586 W	2.734 W

 

I don't know why there would be such differences between the losses of these two means of computation, as the local flux density is the same, we use a NO material, the Bertotti parameters are the same between Flux and Matlab...

Is it coming from flux density values I'm using to compute Matlab losses (I'm using Mod(B)) ? From my Matlab code ?

 

If you need, more precise information is included in joint files to this post.

 

Thank you for your help,

 

Ronan Corin

Explanation Matlab loss computation problem.docx

Ring core issue - programs and FE model.zip

fzidat

Hello Ronan;

 

I am already discussing with you about this request from our professionnal support. 

 

To share with the community, i think that the differences are due to the computation of the barycenter. I sent you a pdf file on how to compute it.

I sent you again your project with mapped mesh, please test it again and make the same comparison, we can see how the mesh quality can have an impact on losses computation, so if the method you use to compute the barycenter is ok, the losses should not change.

 

I asked you also a question, please tell us why you want to replicate the losses computed from Flux with the same model, why to not use Flux directly in this case ?

 

Regards