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.

Boundary Condition on coils cut into 1/4

Please login with a confirmed email address before reporting spam

Hi,

I am simulating 4 circular multiturn coils, and the simulation works correctly when the coils are full loops. However, I want to simplify my geometry by cutting it into 1/4. I cut the geometry into 1/4, and my coils no longer do anything. I know that this is because they are no longer full loops; however, I am trying to use the boundary coniditions so that they are still considered to be full loops. I have tried the magnetic insulation boundary conidition and the perfect magnetic conductor boundary conidition and it still does not work. I am new to comsol so I am not sure if I am using the correct boundary coniditions or if I am using them correctly. I also tried adding a symmetry plane, but that is not even an option for me in my simulation.

I need these coils cut into 1/4 because once I get this working, I am going to add in my magnetic shield also cut into 1/4. The magnetic shield is already in the simulation, but it is just disabled for now. I tried running the simulation with the full loop coils and the full magnetic shield, and I would either get an error talking about not enough space on the RAM or the computation would just freeze and never complete. I believe that this is because the geometry is just too large/ has too many degrees of freedom and that I need to simplify it. Which is why I am trying to cut it into 1/4 and utilize boundary coniditions to get it working. When I cut it into 1/4, the simulation runs, but the coils are not doing what they are supposed to do.

I have attached the file below, if someone could please take a look! I cleared the mesh and the solutions before saving and uploading it onto here, so you will have to mesh it with "extra fine" or else there will be errors. Right now it is already cut into 1/4, but deleting the "delete entities 4,5, and 6" will bring the coils back to full loops.

Thanks!



13 Replies Last Post Oct 20, 2022, 10:04 p.m. EDT

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Aug 23, 2022, 8:35 p.m. EDT
Updated: 2 years ago Aug 23, 2022, 8:36 p.m. EDT

I noticed that I had "form assembly" enabled rather than "form union" and therefore the entire 1/4 of the coils were highlighting when I clicked on the magnetic insulation, so I changed it to "form union" and noticed that now only the boundaries that I wanted to be highlighted were highlighted, yet this still does not fix my problem. When I run the simulation, I still do not see anything happening in the coils. Please let me know if anyone knows what is wrong.

I noticed that I had "form assembly" enabled rather than "form union" and therefore the entire 1/4 of the coils were highlighting when I clicked on the magnetic insulation, so I changed it to "form union" and noticed that now only the boundaries that I wanted to be highlighted were highlighted, yet this still does not fix my problem. When I run the simulation, I still do not see anything happening in the coils. Please let me know if anyone knows what is wrong.

Sergei Yushanov Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Aug 24, 2022, 12:05 p.m. EDT

Victoria,

I made several changes and attached file is working now.

Sergei

Victoria, I made several changes and attached file is working now. Sergei


Please login with a confirmed email address before reporting spam

Posted: 2 years ago Aug 26, 2022, 3:28 p.m. EDT

Victoria,

I made several changes and attached file is working now.

Sergei

Hi Sergei,

Thank you so much for the response. I didn't have COMSOl 6 (I had 5.6) so I had to update it; however, I can now see your solution. I do have a question though. I am not seeing anything plotted when I plot the magnetic field. I added a Cut line 2D and made it from (0,2) to (0,-2) so if you were to imagine the 1/4 coils as full coils, the line would be going through the center of them. Then I added a 1D plot group with my study as the dataset, and then added a line graph with the cut line as that dataset, plotting the magnetic field on the y and the arc length on the x. There should be a magnetic field there; however, nothing comes up. The plot is completely empty, and I am not sure why. I have this exact coil design in 2D, so I know what the field should look like.

Also, I opened my file that I have where the coils are full loops, and I computed that to see the magnetic field again, and once again nothing shows up. However, I have run this exact simulation in the past like two weeks ago (I did not make any changes), and I saw a magnetic field. I even have a screenshot of the plot from the 3D full coils in my notes. But now, I opened that same file back up, reran it, and looked at the plot, and nothing is there. I find COMSOL to be very glitchy in this way. Sometimes it will work and sometimes it won't even if you have changed literally nothing in between runs. I have also attached the file "3Dcoilfull" where I am experiencing this issue.

Basically, on both the 1/4 coil and my full coil, I cannot see the magnetic field and I cannot figure out what is different than when I did it before and DID see the magnetic field.

If you could please let me know if you know how to fix this, I would greatly appreciate it! Thank you so much again

>Victoria, > >I made several changes and attached file is working now. > >Sergei Hi Sergei, Thank you so much for the response. I didn't have COMSOl 6 (I had 5.6) so I had to update it; however, I can now see your solution. I do have a question though. I am not seeing anything plotted when I plot the magnetic field. I added a Cut line 2D and made it from (0,2) to (0,-2) so if you were to imagine the 1/4 coils as full coils, the line would be going through the center of them. Then I added a 1D plot group with my study as the dataset, and then added a line graph with the cut line as that dataset, plotting the magnetic field on the y and the arc length on the x. There should be a magnetic field there; however, nothing comes up. The plot is completely empty, and I am not sure why. I have this exact coil design in 2D, so I know what the field should look like. Also, I opened my file that I have where the coils are full loops, and I computed that to see the magnetic field again, and once again nothing shows up. However, I have run this exact simulation in the past like two weeks ago (I did not make any changes), and I saw a magnetic field. I even have a screenshot of the plot from the 3D full coils in my notes. But now, I opened that same file back up, reran it, and looked at the plot, and nothing is there. I find COMSOL to be very glitchy in this way. Sometimes it will work and sometimes it won't even if you have changed literally nothing in between runs. I have also attached the file "3Dcoilfull" where I am experiencing this issue. Basically, on both the 1/4 coil and my full coil, I cannot see the magnetic field and I cannot figure out what is different than when I did it before and DID see the magnetic field. If you could please let me know if you know how to fix this, I would greatly appreciate it! Thank you so much again


Sergei Yushanov Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Aug 29, 2022, 12:21 p.m. EDT

Victoria,

You have to use "Cut Line 3D" option to plot field distribution along the line in the 3D gerometry. Attached is model with the plot along the specified line.

Sergei

Victoria, You have to use "Cut Line 3D" option to plot field distribution along the line in the 3D gerometry. Attached is model with the plot along the specified line. Sergei


Please login with a confirmed email address before reporting spam

Posted: 2 years ago Aug 31, 2022, 6:31 p.m. EDT

Victoria,

You have to use "Cut Line 3D" option to plot field distribution along the line in the 3D gerometry. Attached is model with the plot along the specified line.

Sergei

Hi Sergei,

Thanks for your help, I see what I was doing wrong there. I have a few more questions, and I hope these will be the last!

First, how were you able to add all of those dependent variables? When I right click on dependant varibles, I do not need an option to add any. Also how did you know that you needed to all of those dependant variables along with the "Gauge fixing for A-field" to get it to work? I am pretty new with comsol, so I am still trying to learn what needs to be in place for certain geometries.

Also, I have now tried to add my magnetic shield in. It was disabled, so I enabled it, and deleted the parts of it that are outside of the quadrant that I am working in. I then enabled the material (silicon steel), and I enabled the Ampere's law 2 because I want the magnetization model for the magnetic shield to be based off of the B-H curve. I meshed it just fine; however, when I go to compute, it gets stuck around 54%. This has happened to me in the past too, it's always 54%. It ends up never computing, or it throws an error. I am not sure if there is supposed to be another boundary condition in there or what. Please let me know if you know how to fix this! The file is attached. This is the final stage that I need my simulation to be at!

Thank you so much for your help.

Best, Victoria

>Victoria, > >You have to use "Cut Line 3D" option to plot field distribution along the line in the 3D gerometry. Attached is model with the plot along the specified line. > >Sergei Hi Sergei, Thanks for your help, I see what I was doing wrong there. I have a few more questions, and I hope these will be the last! First, how were you able to add all of those dependent variables? When I right click on dependant varibles, I do not need an option to add any. Also how did you know that you needed to all of those dependant variables along with the "Gauge fixing for A-field" to get it to work? I am pretty new with comsol, so I am still trying to learn what needs to be in place for certain geometries. Also, I have now tried to add my magnetic shield in. It was disabled, so I enabled it, and deleted the parts of it that are outside of the quadrant that I am working in. I then enabled the material (silicon steel), and I enabled the Ampere's law 2 because I want the magnetization model for the magnetic shield to be based off of the B-H curve. I meshed it just fine; however, when I go to compute, it gets stuck around 54%. This has happened to me in the past too, it's always 54%. It ends up never computing, or it throws an error. I am not sure if there is supposed to be another boundary condition in there or what. Please let me know if you know how to fix this! The file is attached. This is the final stage that I need my simulation to be at! Thank you so much for your help. Best, Victoria


Sergei Yushanov Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Sep 1, 2022, 1:25 p.m. EDT

Victoria,

Modeling magnetic shielding as a thin solid domain is computationally quite expensive (a lot of mesh elements). More efficient way is to use appropriate boundary condition, which is done in the attached model with converged solution. I replaced your 3D solid domain with shell surface and used Magnetic Shielding BCs. Note, that material for shielding is defined on Boundary geometric enttity level (default is Domain). I also changed default solver setting to get convergence.

I admit that coils and inductors is most difficult part of modeling with AC/DC module. There are a lot of resources to learn from: application library examples, lecture series, and various training classes.

Sergei

Victoria, Modeling magnetic shielding as a thin solid domain is computationally quite expensive (a lot of mesh elements). More efficient way is to use appropriate boundary condition, which is done in the attached model with converged solution. I replaced your 3D solid domain with shell surface and used Magnetic Shielding BCs. Note, that material for shielding is defined on Boundary geometric enttity level (default is Domain). I also changed default solver setting to get convergence. I admit that coils and inductors is most difficult part of modeling with AC/DC module. There are a lot of resources to learn from: application library examples, lecture series, and various training classes. Sergei


Please login with a confirmed email address before reporting spam

Posted: 2 years ago Sep 4, 2022, 5:26 a.m. EDT

Plotting field distribution along 3-D gerometry lines requires "Cut Line 3D"Below is the plot along the line specified in the model.

Plotting field distribution along 3-D gerometry lines requires "Cut Line 3D"Below is the plot along the line specified in the model[.](https://maxifoot.info/)

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Sep 6, 2022, 2:11 p.m. EDT

Victoria,

Modeling magnetic shielding as a thin solid domain is computationally quite expensive (a lot of mesh elements). More efficient way is to use appropriate boundary condition, which is done in the attached model with converged solution. I replaced your 3D solid domain with shell surface and used Magnetic Shielding BCs. Note, that material for shielding is defined on Boundary geometric enttity level (default is Domain). I also changed default solver setting to get convergence.

I admit that coils and inductors is most difficult part of modeling with AC/DC module. There are a lot of resources to learn from: application library examples, lecture series, and various training classes.

Sergei

Thank you so much!

>Victoria, > >Modeling magnetic shielding as a thin solid domain is computationally quite expensive (a lot of mesh elements). More efficient way is to use appropriate boundary condition, which is done in the attached model with converged solution. I replaced your 3D solid domain with shell surface and used Magnetic Shielding BCs. Note, that material for shielding is defined on Boundary geometric enttity level (default is Domain). I also changed default solver setting to get convergence. > >I admit that coils and inductors is most difficult part of modeling with AC/DC module. There are a lot of resources to learn from: application library examples, lecture series, and various training classes. > >Sergei Thank you so much!

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Sep 13, 2022, 8:28 p.m. EDT

Victoria,

Modeling magnetic shielding as a thin solid domain is computationally quite expensive (a lot of mesh elements). More efficient way is to use appropriate boundary condition, which is done in the attached model with converged solution. I replaced your 3D solid domain with shell surface and used Magnetic Shielding BCs. Note, that material for shielding is defined on Boundary geometric enttity level (default is Domain). I also changed default solver setting to get convergence.

I admit that coils and inductors is most difficult part of modeling with AC/DC module. There are a lot of resources to learn from: application library examples, lecture series, and various training classes.

Sergei

Hi Sergei,

I just realized that in the new 3D magnetic shield simulation, you only used one Ampere's law. In my 2D magnetic shield simulation, I added a second Ampere's law just for my magnetic shield because I want the "magnetiziation model" of the shield to be based off of the B-H curve, not the relative permability. Would adding this 2nd ampere's law and changing the model to "B-H curve" change anything? I am trying to now add this Ampere's law back and just select the magnetic shield; however, it is not allowing me to do so. It selects other boundaries along with the magnetic shield that I do not want to click. It does not allow me to soley click the magnetic shield. Please let me know! Thanks.

>Victoria, > >Modeling magnetic shielding as a thin solid domain is computationally quite expensive (a lot of mesh elements). More efficient way is to use appropriate boundary condition, which is done in the attached model with converged solution. I replaced your 3D solid domain with shell surface and used Magnetic Shielding BCs. Note, that material for shielding is defined on Boundary geometric enttity level (default is Domain). I also changed default solver setting to get convergence. > >I admit that coils and inductors is most difficult part of modeling with AC/DC module. There are a lot of resources to learn from: application library examples, lecture series, and various training classes. > >Sergei Hi Sergei, I just realized that in the new 3D magnetic shield simulation, you only used one Ampere's law. In my 2D magnetic shield simulation, I added a second Ampere's law just for my magnetic shield because I want the "magnetiziation model" of the shield to be based off of the B-H curve, not the relative permability. Would adding this 2nd ampere's law and changing the model to "B-H curve" change anything? I am trying to now add this Ampere's law back and just select the magnetic shield; however, it is not allowing me to do so. It selects other boundaries along with the magnetic shield that I do not want to click. It does not allow me to soley click the magnetic shield. Please let me know! Thanks.

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Sep 14, 2022, 2:17 a.m. EDT

I suspect you are inadvertently selecting the Air domain surrounding your other domains. If you have the Air domain hidden, it may look like you are selecting all of the boundaries that the Air domain is in contact with.

In general, if your magnetic shield is thin compared to other dimensions in your model, it will be much more computationally efficient to use the magnetic shield BC, rather than an Amperes Law...

Mark

I suspect you are inadvertently selecting the Air domain surrounding your other domains. If you have the Air domain hidden, it may look like you are selecting all of the boundaries that the Air domain is in contact with. In general, if your magnetic shield is thin compared to other dimensions in your model, it will be much more computationally efficient to use the magnetic shield BC, rather than an Amperes Law... Mark

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Sep 15, 2022, 11:46 a.m. EDT

I suspect you are inadvertently selecting the Air domain surrounding your other domains. If you have the Air domain hidden, it may look like you are selecting all of the boundaries that the Air domain is in contact with.

In general, if your magnetic shield is thin compared to other dimensions in your model, it will be much more computationally efficient to use the magnetic shield BC, rather than an Amperes Law...

Mark

Okay, thanks!

>I suspect you are inadvertently selecting the Air domain surrounding your other domains. If you have the Air domain hidden, it may look like you are selecting all of the boundaries that the Air domain is in contact with. > >In general, if your magnetic shield is thin compared to other dimensions in your model, it will be much more computationally efficient to use the magnetic shield BC, rather than an Amperes Law... > >Mark Okay, thanks!

Sergei Yushanov Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Sep 15, 2022, 11:53 a.m. EDT

Victoria,

If you look at my last posted model, you will notice that Magnetic Shielding 1 and Magnetic Shielding 2 boundary conditions actually use B-H curve for the shielding material. The B-H curve is taken from material "Silicon Steel GO 3408" which is defined on the boundary (not on domain as in default case). So, you don't need Ampere's law back (besides, Ampere Law node is defined on the domain level, so even though you put it back, you wouldn't be able to select shield boundary).

Sergei

Victoria, If you look at my last posted model, you will notice that Magnetic Shielding 1 and Magnetic Shielding 2 boundary conditions actually use B-H curve for the shielding material. The B-H curve is taken from material "Silicon Steel GO 3408" which is defined on the boundary (not on domain as in default case). So, you don't need Ampere's law back (besides, Ampere Law node is defined on the domain level, so even though you put it back, you wouldn't be able to select shield boundary). Sergei

Please login with a confirmed email address before reporting spam

Posted: 2 years ago Oct 20, 2022, 10:04 p.m. EDT

Thanks. I have another question regarding the magnetic shield. You had previously made the magnetic shield into a shell because modeling it as a thin domain invovles many mesh elements. However, what if the thickness of the magnetic shield matters? These magnetic shield simulations are supposed to be as realistic as possible, because we are going to be actually use this magnetic shield in real life. The magnetic shield will have a thickness, and by turning it into a shell, we are taking that away and therefore making the simulation less realistic. I have also now added another outer magnetic shield, which will have a different thickness than the inner magnetic shield. However, both of these are currently just being ran as shells. How much of a difference will this make? Do you have any suggestions on how to get the simulation to run without turning the magnetic shield into a shell? Recall that before turning it into a shell, I kept getting errors when trying to run the simulation. Please let me know. Thank you!

Thanks. I have another question regarding the magnetic shield. You had previously made the magnetic shield into a shell because modeling it as a thin domain invovles many mesh elements. However, what if the thickness of the magnetic shield matters? These magnetic shield simulations are supposed to be as realistic as possible, because we are going to be actually use this magnetic shield in real life. The magnetic shield will have a thickness, and by turning it into a shell, we are taking that away and therefore making the simulation less realistic. I have also now added another outer magnetic shield, which will have a different thickness than the inner magnetic shield. However, both of these are currently just being ran as shells. How much of a difference will this make? Do you have any suggestions on how to get the simulation to run without turning the magnetic shield into a shell? Recall that before turning it into a shell, I kept getting errors when trying to run the simulation. Please let me know. Thank you!

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.