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.
Issues of stranded conductors modelling
Posted 14 nov 2009, 21:00 GMT-5 4 Replies
Please login with a confirmed email address before reporting spam
Dear Colleagues,
could you help me to clarify the following situation, please?
I would like to get an advice regarding modeling of stranded conductors. I am using AC/DC module of COMSOL Multiphysics version 3.5a (with hotfix 1, i. e. 3.5.0.608) and my 3D problem defined as follows. Also, feel free to find all the models (2D and 3D for stranded and normal conductors) available in the archive "coils.tar.gz" here -- www.megaupload.com/?d=NNS1OHWD as well.
The domains are:
1.) Air represented as a sphere with a radius of 100 mm;
2.) Coil represented as two subtracted cylinders with a height of 10 mm and winding width 2 mm.
Targets are: magnetic field, electric field and eddy currents induced in a plate (in the future) in these domains.
That is all, just an ordinary time-harmonic time-dependent (duration 1 ms, step 1/4 ms) simulation at a frequency of 1 kHz with a current of 1 A.
It is working well without gauge fixing (without complex numbers allowed both in 2D and 3D) and simulation time is quite fast (0.7 hours for ~34k DOF) while conductivity of the air domain is set to 0 S/m and conductivity of the coil, respectively, to 59 MS/m. Magnetic and electric fields in this case for 2D and 3D models are equal. However, when I am trying to create a stranded conductor and turn the skin effect off by setting the conductivity of the coil to 0 S/m (0.1, 0.01, etc) and turn the gauge fixing on as prescribed in the documentation, it refuses to work! Either "solution is not converged", or "divergence of the linear iterations", even "complex number encountered" (this can be handled through solver settings). By the way, in two dimensional case it works perfect with straight zero conductivities...
I have tried to play with different solvers settings, various conductivity values, gauge fixing, stationary solution, I have tried to include some "dummy" copper plates into the solution... Nothing helps. Is something else wrong with my model? (See the archive, please.)
Although, it works when conductivity of the coil is set to 1 S/m, conductivity of the air is 0 S/m and gauge fixing is on, but it runs very slow (7.4 hours for ~41k) compared to the case where the skin effect is present -- how is that possible? I thought it should be faster because eddy currents are not calculated in the case of no skin effect. Also, magnetic field is comparable, but electric field is not comparable between 2D and 3D models! (Complex numbers were allowed in 2D.) So, will a value of 1 S/m be a good choice for me? Even if I need to include objects with conductivities varied from 0.1 to 0.0001 S/m and an iron core to my solution in the future, will it have any influence onto accuracy of modelling results? (Is the main point here that the value of air conductivity equals to zero?)
Please, advise me how to proceed with the modeling of stranded conductors in COMSOL. What is the proper way to do it regarding electric field and solution time?
I would very appreciate any help. Thank you very much in advance!
With best wishes,
Dmitry
P. S. I would like to thank the COMSOL group for their brilliant product! Eager to try out the fourth version! :)
could you help me to clarify the following situation, please?
I would like to get an advice regarding modeling of stranded conductors. I am using AC/DC module of COMSOL Multiphysics version 3.5a (with hotfix 1, i. e. 3.5.0.608) and my 3D problem defined as follows. Also, feel free to find all the models (2D and 3D for stranded and normal conductors) available in the archive "coils.tar.gz" here -- www.megaupload.com/?d=NNS1OHWD as well.
The domains are:
1.) Air represented as a sphere with a radius of 100 mm;
2.) Coil represented as two subtracted cylinders with a height of 10 mm and winding width 2 mm.
Targets are: magnetic field, electric field and eddy currents induced in a plate (in the future) in these domains.
That is all, just an ordinary time-harmonic time-dependent (duration 1 ms, step 1/4 ms) simulation at a frequency of 1 kHz with a current of 1 A.
It is working well without gauge fixing (without complex numbers allowed both in 2D and 3D) and simulation time is quite fast (0.7 hours for ~34k DOF) while conductivity of the air domain is set to 0 S/m and conductivity of the coil, respectively, to 59 MS/m. Magnetic and electric fields in this case for 2D and 3D models are equal. However, when I am trying to create a stranded conductor and turn the skin effect off by setting the conductivity of the coil to 0 S/m (0.1, 0.01, etc) and turn the gauge fixing on as prescribed in the documentation, it refuses to work! Either "solution is not converged", or "divergence of the linear iterations", even "complex number encountered" (this can be handled through solver settings). By the way, in two dimensional case it works perfect with straight zero conductivities...
I have tried to play with different solvers settings, various conductivity values, gauge fixing, stationary solution, I have tried to include some "dummy" copper plates into the solution... Nothing helps. Is something else wrong with my model? (See the archive, please.)
Although, it works when conductivity of the coil is set to 1 S/m, conductivity of the air is 0 S/m and gauge fixing is on, but it runs very slow (7.4 hours for ~41k) compared to the case where the skin effect is present -- how is that possible? I thought it should be faster because eddy currents are not calculated in the case of no skin effect. Also, magnetic field is comparable, but electric field is not comparable between 2D and 3D models! (Complex numbers were allowed in 2D.) So, will a value of 1 S/m be a good choice for me? Even if I need to include objects with conductivities varied from 0.1 to 0.0001 S/m and an iron core to my solution in the future, will it have any influence onto accuracy of modelling results? (Is the main point here that the value of air conductivity equals to zero?)
Please, advise me how to proceed with the modeling of stranded conductors in COMSOL. What is the proper way to do it regarding electric field and solution time?
I would very appreciate any help. Thank you very much in advance!
With best wishes,
Dmitry
P. S. I would like to thank the COMSOL group for their brilliant product! Eager to try out the fourth version! :)
4 Replies Last Post 6 giu 2017, 12:53 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Dimitry,
Please have a look at my notes attached to see whether there
is anything of use to you. Unfortunately, I am very bussy this
week and I need to have a closer look to your model to give
you more concrete information. Please let me know how it goes.
Bye, Domenico.
Please have a look at my notes attached to see whether there
is anything of use to you. Unfortunately, I am very bussy this
week and I need to have a closer look to your model to give
you more concrete information. Please let me know how it goes.
Bye, Domenico.
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Domenico,
many thanks to you for your help!
I tried to do something the other day, but it did not help me -- unfortunately, I still have this problem.
But I have an idea: could it be that the stranded conductors with zero conductivity are solved fast and, most important, give us the correct electric field only in transient mode instead of time-harmonic? Although, my 2D models are working quite well in the time-harmonic mode, and a section of the user's guide related to the skin effect issues represents the work in 2D with the time-harmonic mode -- do not understand...
I warmly wish you much success in your work. If you have a minute or two, would you be so kind to take a look at the model, please :)
Once again, thank you, Domenico. Have a nice day!
With best wishes,
Dmitry
many thanks to you for your help!
I tried to do something the other day, but it did not help me -- unfortunately, I still have this problem.
But I have an idea: could it be that the stranded conductors with zero conductivity are solved fast and, most important, give us the correct electric field only in transient mode instead of time-harmonic? Although, my 2D models are working quite well in the time-harmonic mode, and a section of the user's guide related to the skin effect issues represents the work in 2D with the time-harmonic mode -- do not understand...
I warmly wish you much success in your work. If you have a minute or two, would you be so kind to take a look at the model, please :)
Once again, thank you, Domenico. Have a nice day!
With best wishes,
Dmitry
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Dmitry,
Stranded conductors should indeed have zero conductivity.
Best, Domenico.
Stranded conductors should indeed have zero conductivity.
Best, Domenico.
Please login with a confirmed email address before reporting spam
Posted:
7 years ago
Hello Dimitry,
Your problem is interesting and the same nature issue I am facing while modeling stranded conductors. My task is to couple heat transfer in solids to determine the temperature coefficients used for different materials in conductors.
Recently I read an excellent tutorial on submarine cables from COMSOL which is available in the application library, unfortunately, it only works with version 5.3. Due to the contrast in conductivity, there could be a problem to find the solution. I wonder if you are using insulation system in your model? If yes, that is the case of "ill-posed problem".
If you achieve any progress in the model, I would greatly appreciate if you can share details with me.
Thank you,
Nirav
Your problem is interesting and the same nature issue I am facing while modeling stranded conductors. My task is to couple heat transfer in solids to determine the temperature coefficients used for different materials in conductors.
Recently I read an excellent tutorial on submarine cables from COMSOL which is available in the application library, unfortunately, it only works with version 5.3. Due to the contrast in conductivity, there could be a problem to find the solution. I wonder if you are using insulation system in your model? If yes, that is the case of "ill-posed problem".
If you achieve any progress in the model, I would greatly appreciate if you can share details with me.
Thank you,
Nirav