Discussion Forum

In addition, both results are more than 1 order of magnitude lower than the results I get from Ansoft Maxwell. In Ansoft Maxwell, I used exactly the same boundary and excitation conditions and I get the center B to be 2T instead of 0.07T I get in comsol. Could someone tell me why? Thank you very much!

Hi

Are you sure you have "exactly the same" boundary and model conditions ?

my comments are first your very limited air in your model does not allow the field to loop around correctly, you strongly influences the behaviour of your model, you could also add some infinite elements on the outer side, and a symmetry on the radial line z=0 to improve things

Then I cannot understand why you get about 1:1000 difference between the two softwares. What would you expect for B by a quick analytical calculation ?

And why not start with linear core just to see if the model agrees with analytical values ?

Another possibility, do you need a mur defined or not ? how are your B-H curves normalised ?

--
Good luck
Ivar

Hi, Ivar,
Thank you very much for your help. I changed the air region and I do get a much better and reasonable result compared to Ansoft Maxwell. However, I still don't quite understand why my two simulations using different material with other boundary conditions exactly the same are so close to each other. There's slight different in Magnetization however when I plot B curve the difference is unnoticeable.
I'm concerned if BH curve is actually used in the computation processing since the BH curve for the two materials are complete different. Thank you. Any advice will be very much appreciated.

--
Charles

Hi

that is good to hear ;) the two software should agree, for similar models

I find the two rather different, try to plot them on the same curve.
And do also consider the size effect, magnetic effects do rapidly loose interest for small sizes, compared to electrostatic force devices.

My feeling, but I would have to run some numebr over it before I can shout it out loud, is that for such small sizes the saturation response is also more sensitive than for larger items. Your model is also using high current densities, 3 to at most 10 A/mm^2 are typical for macroscopical devices, but again it's small so surface to volume effects are different ;)

--
Good luck
Ivar

Thank you.
I actually do have a question about the saturation. Because the embedded material property for iron has the BH curve that doesn't saturate till B reaches around 1T. Since the peak value in the middle of the solenoid is only 0.6T, does it not mean the core material for both cases have not saturated yet? Thanks.


--
Charles

Hi

I agree that it did not look like it was saturating, but again I looked rather quickly ;)
Nrmally 0.6 T in the midle should not imly much higher for the rest of the module (in steady state)

--
Good luck
Ivar

Thank you! So that means saturation doesn't happen in the whole model, right?

--
Charles