Magnetization

The graphical user interface of Elmer
Post Reply
markuss01
Posts: 1
Joined: 07 Sep 2021, 10:57
Antispam: Yes

Magnetization

Post by markuss01 »

Hi,

I am working with Elmer GUI (Windows 10) for 2 weeks to come familiar the first time with 3D-static electromagnetics (for training I did the GUI tutorials and adapted the cases for my application) and have some questions:

1. I can´t find in different material sources the Magnetization as specific property. Instead of this, there are often values for the coercivity bHc or intrinsic coercivity iHc in [A/m]
https://www.magnetladen.de/media/pdf/a4 ... 5MxtMV.pdf

I am using iHc for the Magnetization material Parameter in Elmer because of B=µ*(H+M). Is it right?

2. I am not sure about the convergence on mgdynpost, look at the attachment. I am using the standard parameters like mgdyn with BiCGstabl and no preconditioner. The results in Paraview are looking fine.
convergence.png
convergence.png (160.53 KiB) Viewed 2996 times
solveroutput.txt
(18.9 KiB) Downloaded 164 times
case.sif
(4.46 KiB) Downloaded 180 times
3. How do I activate the use of quadratic elemens in Elmer gui. Manipulating the sif-file is not elegant, because of changing its content, when you start a new simulation.

I am looking forward to hear from you.
Markus :)
raback
Site Admin
Posts: 4801
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Magnetization

Post by raback »

Hi

1) Maybe somebody else can comment on this...

2) No need to worry here. The Post solves different fields using galerkin method and each field has different norm. So the convergence plot does not really make sense here.

3) Yes, it seems that the "quadratic approximation = true" keyword is missing from the magnetodynamics.xml file. You can use "free text input" or go and edit the xml file to add the keyword. It would make sense to have it there as the solver can deal with 2nd order edge elements.

-Peter
Rich_B
Posts: 421
Joined: 24 Aug 2009, 20:18

Re: Magnetization

Post by Rich_B »

Hi,

Regarding question #3, you can tell elmergrid the command '-increase' within ElmerGUI, following these steps:

viewtopic.php?f=3&t=7610&p=25772&hilit= ... rid#p25772

except, instead of adding '-merge 1.0e-5', add '-increase', then remesh. This will increase the elements from linear to quadratic, and will write the results into the Elmer *.mesh files. You shouldn't need to make any changes to your sif file.

Please let us know how this works for you.

Rich.
raback
Site Admin
Posts: 4801
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Magnetization

Post by raback »

Rich, I think he is using edge elements which are quite different from the nodal elements that may be used in most other modules. To be more specific, the scalar potential is given in nodes but the vector potential that is here more relevant is solved using the edge element basis.

-Peter
Rich_B
Posts: 421
Joined: 24 Aug 2009, 20:18

Re: Magnetization

Post by Rich_B »

Hi Peter,

Welcome back, hope you had a great summer.

Sorry about mixing up nodal elements and edge elements, it's not a topic that I'm familiar with.

So edge elements are created by the solver using the elmer mesh as a 'background mesh'. Increasing the order of the edge elements can only be done by the solver, so the commands have to go into the sif file. The WhitneyAVsolver is the only solver that uses edge elements, correct?

Thanks, Rich.
mika
Posts: 230
Joined: 15 Sep 2009, 07:44

Re: Magnetization

Post by mika »

Rich_B wrote: 08 Sep 2021, 15:45 The WhitneyAVsolver is the only solver that uses edge elements, correct?
The modules EMWaveSolver and VectorHelmholtz for electromagnetic waves also use edge finite elements.

-- Mika
Rich_B
Posts: 421
Joined: 24 Aug 2009, 20:18

Re: Magnetization

Post by Rich_B »

Okay, all three of those solvers have the keyword 'Use Piola Transform'.

Is there a guideline when to increase the edge elements for these solvers?

I'm asking because it might make sense to add some details about increasing edge elements to the tutorials. If we wanted to add a paragraph or two, what should be included?

Thanks, Rich.
raback
Site Admin
Posts: 4801
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Magnetization

Post by raback »

Hi Rich,

Adding polynomial degree to elements is a double-edged sword. Basically it of course increases accuracy but at the same time the computational load in setting up and solving the system grows significantly. Also the resulting linear system is more difficult to solve for iterative solvers. It is really difficult to give strict guidelines. Usually it is good to start with linear elements because they are more economical and robust. There are problems that rely heavily on derivatives where the added accuracy is particularly useful. For example if you want to compute heat fluxes or magnetic nodal forces you might be better off using higher order elements.

-Peter
Post Reply