Menu

Discussion Forum

New Discussion
Filtra

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.

Electromagnetic Force X,Y and Z Component

Posted 9 ott 2014, 03:39 GMT-4 RF & Microwave Engineering Version 5.1 19 Replies

Thomas Bender
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Hi@All,


I have created a simulation of a metal ball in an electromagnetic field of a copper coil.
Unfortunately, the forces are that I have not determined correctly (at least I think they are wrong ;) ).
In my deliberation should the force in the X-axis have the biggest value and Y and Z components would be negligible.
It would be great if someone in the forum could look at the simulation and may be provide with some advice.
It would also be very nice if someone can give me a tip on how I interpret the ball as Magnetic ball (magnetic-axis = x-axis).

Thanks and best regards Thomas
19 Replies Last Post 1 ago 2017, 14:08 GMT-4
Sergei Yushanov Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 9 ott 2014, 09:24 GMT-4
Thomas,

A few notes about your model:

1. Your mesh for ball is too coarse. With such coarse mesh electromagnetic force is just numerical noise. Force calculation node is using integration of the Maxwell stress tensor over the surface of body to evaluate the force components in x-, y-, and z-directions. Fine and good quality mesh is required to get accurate results for the electromagnetic force.

2. Your coil is too close to the boundary of the computational domain. Increase air domain radius or use “Infinite Element Domain” domain feature available under Model/Definitions.

3. To make the ball magnetic, right-click Magnetic Fields (mf) node and select “Ampere’s Law”. This will add new “Ampere’s Law 2” node. Go to the Settings Window for Ampere’s Law and in the Domain Selection section select domain 4 (this is ball domain). In the Magnetic Field/Constitutive relation section, select “Remanent flux density” and define the magnitude of the remanent flux density in the x-direction. This will make the ball a permanent magnet with magnetization along the x-axis.

4. You problem is axisymmetric. I would suggest starting with 2D axisymmetric geometry to solve the problem. Once you are happy with results you can move on and solve the same problem in 3D geometry.

Regards,
Sergei
Thomas, A few notes about your model: 1. Your mesh for ball is too coarse. With such coarse mesh electromagnetic force is just numerical noise. Force calculation node is using integration of the Maxwell stress tensor over the surface of body to evaluate the force components in x-, y-, and z-directions. Fine and good quality mesh is required to get accurate results for the electromagnetic force. 2. Your coil is too close to the boundary of the computational domain. Increase air domain radius or use “Infinite Element Domain” domain feature available under Model/Definitions. 3. To make the ball magnetic, right-click Magnetic Fields (mf) node and select “Ampere’s Law”. This will add new “Ampere’s Law 2” node. Go to the Settings Window for Ampere’s Law and in the Domain Selection section select domain 4 (this is ball domain). In the Magnetic Field/Constitutive relation section, select “Remanent flux density” and define the magnitude of the remanent flux density in the x-direction. This will make the ball a permanent magnet with magnetization along the x-axis. 4. You problem is axisymmetric. I would suggest starting with 2D axisymmetric geometry to solve the problem. Once you are happy with results you can move on and solve the same problem in 3D geometry. Regards, Sergei
Thomas Bender
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 ott 2014, 03:44 GMT-4
Hello Sergei,

thank you for your quick and detailed answer.
1+2: I increases the radius of the air domain for the factor of 10 and I have made a new mesh with an extremely fine resolution for the ball, but it remains in the curious result of the forces (the X-component is less than the Y component and Z).

3: This method worked very well, thank you.

4: I will try your suggestion today in a new simulation and will publish the results here ;)

Another point occurred as I compute the current in the coil with an Electrical Circuit. I have created a circuit with a "Ground Node", "Voltage Source", "External I Vs. U" and "Resistor". The simulation takes forever and then simply interrupts with an error :(

Thanks for your help
sincerely Thomas
Hello Sergei, thank you for your quick and detailed answer. 1+2: I increases the radius of the air domain for the factor of 10 and I have made a new mesh with an extremely fine resolution for the ball, but it remains in the curious result of the forces (the X-component is less than the Y component and Z). 3: This method worked very well, thank you. 4: I will try your suggestion today in a new simulation and will publish the results here ;) Another point occurred as I compute the current in the coil with an Electrical Circuit. I have created a circuit with a "Ground Node", "Voltage Source", "External I Vs. U" and "Resistor". The simulation takes forever and then simply interrupts with an error :( Thanks for your help sincerely Thomas
Thomas Bender
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 ott 2014, 05:10 GMT-4
Hello again,

I have created a 2D Axissymmetric model, but unfortunately is also the Z-component (compared to 3D model the X_Komponente) much too small Iit would be very nice if someone in this forum could view my model and give me some hints.

Thanks for your help
sincerely Thomas



Hello again, I have created a 2D Axissymmetric model, but unfortunately is also the Z-component (compared to 3D model the X_Komponente) much too small Iit would be very nice if someone in this forum could view my model and give me some hints. Thanks for your help sincerely Thomas
Sergei Yushanov Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 ott 2014, 09:21 GMT-4
Thomas,

Attached is the test model for electromagnetic force calculation. Model goes slightly beyond the basics and couples the electromagnetic force to Newton law of motion for the ball. You can see animation of ball motion under the node Results/Export/Player. Parameters of the test model are arbitrary and are not related to anything specific.

Regards,
Sergei
Thomas, Attached is the test model for electromagnetic force calculation. Model goes slightly beyond the basics and couples the electromagnetic force to Newton law of motion for the ball. You can see animation of ball motion under the node Results/Export/Player. Parameters of the test model are arbitrary and are not related to anything specific. Regards, Sergei
Thomas Bender
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 13 ott 2014, 05:27 GMT-4
Hello Sergei,

1000 times thank you for your help, the new model is almost exactly what I'm looking for.
I have tried to magnetize the ball with a 2nd Ampère's Law, but unfortunately the movement of the ball is less. However, the motion shall be greater because the ball would attract even more strongly. Unfortunately I can not see the error.
Additionally, I have added a ferrite in the coil, but the force remains the same. Have I done something wrong?

Thanks again for your help,
with best regards
Thomas
Hello Sergei, 1000 times thank you for your help, the new model is almost exactly what I'm looking for. I have tried to magnetize the ball with a 2nd Ampère's Law, but unfortunately the movement of the ball is less. However, the motion shall be greater because the ball would attract even more strongly. Unfortunately I can not see the error. Additionally, I have added a ferrite in the coil, but the force remains the same. Have I done something wrong? Thanks again for your help, with best regards Thomas
Sergei Yushanov Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 13 ott 2014, 10:49 GMT-4
Thomas,

1. In the modified model you set magnetization in the r-direction for the spherical ball with high relative permeability. This seems odd to me – not sure what is physical interpretation of this. I would recommend studying one thing at a time. Set relative permeability to one and use Ampere’s law to model ball as a permanent magnet. Once you are happy with the results, change relative permeability of the ball and observe the effect. Attached is the model where the ball is modeled as permanent magnet with remanent flux density in the z-direction.

2. In your modified model the selection for Material 4 (mat4) is empty.

3. You added the Coil Core and didn’t change Moving Mesh (ale) setup accordingly. As is, your setup for Moving Mesh interface is completely wrong.

4. Mesh for the modified model is incorrect – mesh of the Coil Core overlays the mapped mesh of the air domain.

5. Ball can’t move through the solid Coil Core.

Regards,
Sergei
Thomas, 1. In the modified model you set magnetization in the r-direction for the spherical ball with high relative permeability. This seems odd to me – not sure what is physical interpretation of this. I would recommend studying one thing at a time. Set relative permeability to one and use Ampere’s law to model ball as a permanent magnet. Once you are happy with the results, change relative permeability of the ball and observe the effect. Attached is the model where the ball is modeled as permanent magnet with remanent flux density in the z-direction. 2. In your modified model the selection for Material 4 (mat4) is empty. 3. You added the Coil Core and didn’t change Moving Mesh (ale) setup accordingly. As is, your setup for Moving Mesh interface is completely wrong. 4. Mesh for the modified model is incorrect – mesh of the Coil Core overlays the mapped mesh of the air domain. 5. Ball can’t move through the solid Coil Core. Regards, Sergei
Thomas Bender
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 17 ott 2014, 09:07 GMT-4
Dear Sergei,
hi@all,

unfortunately I'm still working on the simulation model. I used your file / model as basis and transfer my (small) geometry.
I disabled the coil core and the permanent magnet part of the ball, for a first test. (Disabled the Rectangle for the core and Ampère's Law 2: Permanent Magnet)
Unexpectedly the ball falls down instead of an attraction of the coil, the system also seems very slow.
It would be very nice if you could look and check at my model.

Thanks a lot,
Thomas
Dear Sergei, hi@all, unfortunately I'm still working on the simulation model. I used your file / model as basis and transfer my (small) geometry. I disabled the coil core and the permanent magnet part of the ball, for a first test. (Disabled the Rectangle for the core and Ampère's Law 2: Permanent Magnet) Unexpectedly the ball falls down instead of an attraction of the coil, the system also seems very slow. It would be very nice if you could look and check at my model. Thanks a lot, Thomas
Thomas Bender
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 21 ott 2014, 03:21 GMT-4
Dear Sergei,
hi@all,

my understanding of the model continues to grow. I noticed that the calculation of the "Global Equations 1 Ub" has the wrong unit.

You use Uz as shift path
* with a constant part: t * 3 [m / s] * 0 ( = 0 [m] )
* plus a dynamic part: Ub (Ub = mb * Ubtt - mf.Forcez_ball
* mb = mass of the ball
* Ubtt = acceleration of Ub (second derivation of Ub)
* mf.Forcez_ball = Force form the Magnetic Fields (mf) which act on the ball in the Z-axis

Ub should have the same unit as Uz [m], right?
For the Prescribed Mesh Displacement 3 and 4 do you use the variable "s" or "(1-s)", where "s" is determined?
It would be very nice if someone could look at this part and may give me some tips.

Thanks a lot,
Thomas
Dear Sergei, hi@all, my understanding of the model continues to grow. I noticed that the calculation of the "Global Equations 1 Ub" has the wrong unit. You use Uz as shift path * with a constant part: t * 3 [m / s] * 0 ( = 0 [m] ) * plus a dynamic part: Ub (Ub = mb * Ubtt - mf.Forcez_ball * mb = mass of the ball * Ubtt = acceleration of Ub (second derivation of Ub) * mf.Forcez_ball = Force form the Magnetic Fields (mf) which act on the ball in the Z-axis Ub should have the same unit as Uz [m], right? For the Prescribed Mesh Displacement 3 and 4 do you use the variable "s" or "(1-s)", where "s" is determined? It would be very nice if someone could look at this part and may give me some tips. Thanks a lot, Thomas
Sergei Yushanov Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 ott 2014, 00:15 GMT-4
Thomas,

1. The orange color you in the Global Equations settings window is not because the units are wrong but because the units for the global variable “Ub” are not specified at all. If you specify units of length (m) for the Dependent variable quantity Ub and appropriate units for the Source term quantity the color becomes black as shown in the first attached image.

2. Comsol uses internal variable “s” for the Arc length parameter. Parameter s=0 at the beginning of any arc and s=1 at the end of the arc. You can see this by creating 1D Plot Group/Line Graph, then select any edge and type “s” as Expression for the y-Axis Data. Also, you can visualize the direction of edges by checking box “Show edge direction arrows” in the View node, as shown in the second attached image.

Regards,
Sergei

Thomas, 1. The orange color you in the Global Equations settings window is not because the units are wrong but because the units for the global variable “Ub” are not specified at all. If you specify units of length (m) for the Dependent variable quantity Ub and appropriate units for the Source term quantity the color becomes black as shown in the first attached image. 2. Comsol uses internal variable “s” for the Arc length parameter. Parameter s=0 at the beginning of any arc and s=1 at the end of the arc. You can see this by creating 1D Plot Group/Line Graph, then select any edge and type “s” as Expression for the y-Axis Data. Also, you can visualize the direction of edges by checking box “Show edge direction arrows” in the View node, as shown in the second attached image. Regards, Sergei
Thomas Bender
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 ott 2014, 05:34 GMT-4
Dear Sergei,

thank you for your notes.
I think: (@ Definitionen -> Variables1)
Uz = t * 1[m/s] * 0 + Ub
*with the units [m] = [s] * [m/s] * [ ] + [m]
is correct and (@ Global ODEs and DAEs: Ball Motion -> Global Equations 1)
Ub = mb * Ubtt - Mf.Forcez.Ball
*with the units [m] = [kg] * [m/(s^2)] - [N]
unfortunately, this is not correct, Ub should also have the unit [N] ?
But Comsol shows no Error, why? Where did I make a mistake?

With best regards, Thomas
Dear Sergei, thank you for your notes. I think: (@ Definitionen -> Variables1) Uz = t * 1[m/s] * 0 + Ub *with the units [m] = [s] * [m/s] * [ ] + [m] is correct and (@ Global ODEs and DAEs: Ball Motion -> Global Equations 1) Ub = mb * Ubtt - Mf.Forcez.Ball *with the units [m] = [kg] * [m/(s^2)] - [N] unfortunately, this is not correct, Ub should also have the unit [N] ? But Comsol shows no Error, why? Where did I make a mistake? With best regards, Thomas
Sergei Yushanov Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 ott 2014, 00:44 GMT-4
Thomas,

The first column in the setting window for Global equations is the name of the variable being solved for, i.e. you are solving for variable Ub.
The second column is for declaration of the Left Hand Side of equation you are solving, i.e. LHS=mb*Ubtt-mf.Forcez_ball

The actual equation you are solving is as following:
LHS=0
or
mb*Ubtt-mf.Forcez_ball=0
Both terms of this equation have the same units of force [N]. So, units are consistent and correct, see attached image.

Regards,
Sergei
Thomas, The first column in the setting window for Global equations is the name of the variable being solved for, i.e. you are solving for variable Ub. The second column is for declaration of the Left Hand Side of equation you are solving, i.e. LHS=mb*Ubtt-mf.Forcez_ball The actual equation you are solving is as following: LHS=0 or mb*Ubtt-mf.Forcez_ball=0 Both terms of this equation have the same units of force [N]. So, units are consistent and correct, see attached image. Regards, Sergei
Thomas Bender
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 ott 2014, 04:11 GMT-4
Hi Sergei,

thanks again. So the actually calculation in Comsol is the formula:

Ubtt = mf.Forcez_ball / mb with de purity [m / s ^ 2]
with mf.Forcez_ball from Magnetic Fields (mF) -> Force Calculation 1
and mb fixed by Parameters: rho_b*v_b = 5.1051E-4kg
and from this the way Ub is determined.

thank you and I wish you a nice day
Thomas
Hi Sergei, thanks again. So the actually calculation in Comsol is the formula: Ubtt = mf.Forcez_ball / mb with de purity [m / s ^ 2] with mf.Forcez_ball from Magnetic Fields (mF) -> Force Calculation 1 and mb fixed by Parameters: rho_b*v_b = 5.1051E-4kg and from this the way Ub is determined. thank you and I wish you a nice day Thomas
Nichith Chandrasekaran
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 4 nov 2014, 08:36 GMT-5
Hi Sergei,

I am trying to simulate a similar kind of problem. But the difference is that, in my case the armature keeps moving forward due to the action of EM force. Here the ball is oscillating. So my doubt is in the Moving mesh part of the test file. I see that no Automatic remeshing is used. But if the ball were to keep moving without oscillating then we would require remeshing. How can I use that feature. Because my simulation depends on that. I have attached my file here. It would be very helpful if I could get some suggestion regarding this.

Thanks and regards
Nichith
Hi Sergei, I am trying to simulate a similar kind of problem. But the difference is that, in my case the armature keeps moving forward due to the action of EM force. Here the ball is oscillating. So my doubt is in the Moving mesh part of the test file. I see that no Automatic remeshing is used. But if the ball were to keep moving without oscillating then we would require remeshing. How can I use that feature. Because my simulation depends on that. I have attached my file here. It would be very helpful if I could get some suggestion regarding this. Thanks and regards Nichith

Attachments:

Sergei Yushanov Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 5 nov 2014, 07:42 GMT-5
Nichith,

A few notes on your model:

1. Remeshing algorithm would not allow you to set armature motion over the large distance.

2. Test file uses sliding mesh algorithm to model motion of body over large distances. Your implementation of the sliding mesh is not correct. You need to add Prescribed Mesh Displacement nodes for the sliding edges and use arc length parameter “s” to prescribe edge displacement as a linear function of parameter “s”.

3. There are three domains in you model which are not connected to each other. It means that solution in domains 1 and 7 will be zero.

4. I don’t see the need for adding nodes ”Ampere Law 2” and “Ampere Law 3” in your model.

5. Not sure if you can impose Continuity BC along the moving interface between two conductors (aluminum/copper in your model). Moving conductor will create surface current and interface surface current will introduce field discontinuity across the interface.

Regards,
Sergei
Nichith, A few notes on your model: 1. Remeshing algorithm would not allow you to set armature motion over the large distance. 2. Test file uses sliding mesh algorithm to model motion of body over large distances. Your implementation of the sliding mesh is not correct. You need to add Prescribed Mesh Displacement nodes for the sliding edges and use arc length parameter “s” to prescribe edge displacement as a linear function of parameter “s”. 3. There are three domains in you model which are not connected to each other. It means that solution in domains 1 and 7 will be zero. 4. I don’t see the need for adding nodes ”Ampere Law 2” and “Ampere Law 3” in your model. 5. Not sure if you can impose Continuity BC along the moving interface between two conductors (aluminum/copper in your model). Moving conductor will create surface current and interface surface current will introduce field discontinuity across the interface. Regards, Sergei
Nichith Chandrasekaran
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 6 nov 2014, 16:09 GMT-5
Hi Sergei,

Thank you for the suggestions. I was able to make a few changes to my model. The moving mesh seems to work well now. As for the Continuity BC, I am not sure how to solve this. Because if I don't use the continuity BC, the boundary gets insulated in which case there is no current flowing through the armature. Is there any other way to solve this issue?

Thanks and Regards,
Nichith
Hi Sergei, Thank you for the suggestions. I was able to make a few changes to my model. The moving mesh seems to work well now. As for the Continuity BC, I am not sure how to solve this. Because if I don't use the continuity BC, the boundary gets insulated in which case there is no current flowing through the armature. Is there any other way to solve this issue? Thanks and Regards, Nichith

Attachments:

Sergei Yushanov Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 nov 2014, 08:26 GMT-5

Nichith,

You can solve this problem using Velocity (Lorentz Term) available in Comsol. Then you don’t need Assembly mode and Moving Mesh. I would suggest read carefully documentation on how to use Velocity (Lorentz Term) ferature.

Regards,
Sergei
Nichith, You can solve this problem using Velocity (Lorentz Term) available in Comsol. Then you don’t need Assembly mode and Moving Mesh. I would suggest read carefully documentation on how to use Velocity (Lorentz Term) ferature. Regards, Sergei
Nichith Chandrasekaran
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 10 years ago 29 nov 2014, 07:27 GMT-5
Hi Sergei,

Using the velocity (Lorentz term), the armature will is not moved if the moving mesh is not considered. Because I need to further extend this to complex geometries. And I need to study the movement of the armature. I found a tutorial problem I have attached below(falling magnet). There is no movement of the object.

Also I have tried to change my previously attached file. Now it seems to work. But does it seem right? Can you please give some suggestions.

Thank you
Hi Sergei, Using the velocity (Lorentz term), the armature will is not moved if the moving mesh is not considered. Because I need to further extend this to complex geometries. And I need to study the movement of the armature. I found a tutorial problem I have attached below(falling magnet). There is no movement of the object. Also I have tried to change my previously attached file. Now it seems to work. But does it seem right? Can you please give some suggestions. Thank you
Kumar Vishal
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 21 ago 2015, 12:52 GMT-4
Hi sergei,

I am making RF design using comsol, I am also using comsol from quit a long time now. I finished my design and verified my results but it is totally different from my theoretical results. My device should work like non-reciprocal but I couldn't see any non-reciprocal effect in s21 or s12 parameter. Please have a look at it if you gets time.

I have attached a model and related paper here.

I have some other questions too:

1.I extract the s-parameter and that file have coordinates and s-parameter. My question is how can i make sure s-parameter (s12, s21) which I got directly using comsol is correct or do I need to find the boundary point in that extract file and use mat lab to plot the curve for (s12, s21).

2. Does my result in comsol plot for (s12, s21) shows me the right graph, do I need to use matlab.

3. What should I do to improve my design. I make exactly same design as I made in lab and measure it theoretically.

4. When we try to plot s-parameter using comsol does it plot s-parameter on boundary we selected or some thing different.
Hi sergei, I am making RF design using comsol, I am also using comsol from quit a long time now. I finished my design and verified my results but it is totally different from my theoretical results. My device should work like non-reciprocal but I couldn't see any non-reciprocal effect in s21 or s12 parameter. Please have a look at it if you gets time. I have attached a model and related paper here. I have some other questions too: 1.I extract the s-parameter and that file have coordinates and s-parameter. My question is how can i make sure s-parameter (s12, s21) which I got directly using comsol is correct or do I need to find the boundary point in that extract file and use mat lab to plot the curve for (s12, s21). 2. Does my result in comsol plot for (s12, s21) shows me the right graph, do I need to use matlab. 3. What should I do to improve my design. I make exactly same design as I made in lab and measure it theoretically. 4. When we try to plot s-parameter using comsol does it plot s-parameter on boundary we selected or some thing different.
Mohammad Qasim
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 7 years ago 1 ago 2017, 14:08 GMT-4
ri8. dear, it was good effort. but in the model u are doing little mistakes,

1. ur global equations variables got inconsistent units, try to resolve it by giving dimensions to dependent and independent variables.
2. extend the geometry of rails to make the time convergence.
3. 3d model of the problem, may be more preferable.
ri8. dear, it was good effort. but in the model u are doing little mistakes, 1. ur global equations variables got inconsistent units, try to resolve it by giving dimensions to dependent and independent variables. 2. extend the geometry of rails to make the time convergence. 3. 3d model of the problem, may be more preferable.

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.

Suggested Content