Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
How to apply current in r-z plane in a 2D AxiSym model?
Posted 4 giu 2020, 08:27 GMT-4 Low-Frequency Electromagnetics Version 5.4 7 Replies
Please login with a confirmed email address before reporting spam
Dear COMSOLers,
In my 3D model (prenested in the COMSOL Discussion Forum here) I want to study the skin effect in the conductor body (in frequancy domian study). Because the condcutor thickness is very small and at higher frequencies the current is being driven closer to the boundaries, I need to apply much finer mesh (to the conductor domain) to resolve the spatial variation of the fields. With the structure large aspect ratio, it seems very diffcult for my computational resources to withstand this huge meshing matrix. Since my model is azimuthely symmetric, I decided to study it in a 2D Axial-Symetric reference farme (attached is a simplified model). However, I faced an "old" problem: how I can apply the current in r-z plane instead of normal to plane? I see that this issue was discussed decade ago here, however, I still cannot get how to do it. It also seems that there is a similar study done by COMSOL and published here, but I cannot find details on this study!
How can I apply current in r-z plane in 2D axial-symmetry model?
Regards, Tamer
Attachments:
Please login with a confirmed email address before reporting spam
Tamer,
I think this can be done with a coaxial port between the two coil terminals at the top of the model. No coil node needed then.
Cheers Edgar
-------------------Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Please login with a confirmed email address before reporting spam
Hi Edgar,
The idea of the coaxial port seems the right way in this case, however: 1- Option "Coaxial" of the "Type of lumped port:" is not available for any of the AC/DC module physics nodes!! I am using "Magnetic field (mf)". 2- While it is different physics, I tried to guid myself abit with this example, to understand how the lumped port works. So, I defined the input and output ports (with terminal type = current), please see the attachment. However, this gave me non-understandable results. 3- I also tried to assign the terminal between the two conductors as you suggested, but still the solution is not correct.
Just for guidness, I also attach the corresponding/correct/expected result from the 3D model!
Any advice!
Regards, Tamer
Attachments:
Please login with a confirmed email address before reporting spam
Tamer,
you need to set the mf physics to solve for in-plane vector potential, since you only want to see a phi-component of the magnetic field, right? Then the coaxial port gets available and your Lumped Port 3 is the correct way.
Cheers Edgar
-------------------Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Please login with a confirmed email address before reporting spam
Hi Edgar,
Modifying the input parameters of the mf physics to solve in-plane helped indeed to activate the coaxil option. Now, solving the model without adding any further boundary conditions, than Port 3, gives me the "The relative residual (1.1e+49) is greater than the relative tolerance." error message!! Indication of a huge problem either in the: - defined boudary condition ==> I reduced the applied current to very low level. - or meshing startegy ==> I modifed my meshing to a very efficient boundary layers style. - I also tried to change the direct solver to PARDISO. - I also treid to change the termination process depending on a large number of iterations.
What I am still missing? The modified model is attached.
Regards, Tamer
Attachments:
Please login with a confirmed email address before reporting spam
The solver sequence is not correct for the stationary study. The frequency domain step solves if you assign a small conductivity to the air domain, e.g. 1 S/m. Also you may need, or not, to add a Gauge Fixing node. Compare the results with and without, if the difference is significant you need it. It consumes some more memory.
-------------------Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Please login with a confirmed email address before reporting spam
Dear Edgar,
Many thanks for your guidance throughout this case. It works perfect now.
As usual and in order to close this case in a useful way, I attach the working model as a reference for the future similar cases. I also want to state that the Gauge Fixing node was not making any difference for this specific model. However, as Edgar recommended, always check its infuence for each specific case.
Best regards, Tamer
Attachments:
Please login with a confirmed email address before reporting spam
Tamer,
thanks for the feedback and good luck for your work.
Best regards Edgar
-------------------Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com