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 issues
Posted Aug 6, 2009, 11:53 a.m. EDT 10 Replies
Please login with a confirmed email address before reporting spam
I was given the task of determining the attractive force vs distance between two electromagnets with varied inner iron core geometries. However, upon running COMSOL with my core set as Iron and then again as nothing, I saw no change which leads me to believe I'm missing something in my setting for the iron core I have.
It would be simplest to understand all of this by opening the file I've attached.
First of all, I'm not absolutely positive that I'm using the best Module to be modeling this. (AC/DC > Statics, Magnetic > Magnetostatics, vector potential)
I'm trying to do a trade study but the only information I was given was that a standard copper wire carrying either 12 V (.6 Amps) or 24 V (1.1 Amps) was wrapped around a solid cylindrical iron core (approx 40 turns) to create an electromagnet. There are two of these exact same devices a short distance apart (I've been starting at approx. 0.25 cm) attracting each other due to the parallel current in them. The only information I want is the Attractive Force (N) vs. Axial Distance (cm) of a few different iron core geometries, but I can't seem to get a sensible answer from the Electromagnetic force in the Global Data Display as I said before when I 'turned off' the iron core.
For some reason the iron has no influence... how do I fix this? Is there a better way to get more accurate axial force data?
Any and all suggestions welcome at this point. :)
It would be simplest to understand all of this by opening the file I've attached.
First of all, I'm not absolutely positive that I'm using the best Module to be modeling this. (AC/DC > Statics, Magnetic > Magnetostatics, vector potential)
I'm trying to do a trade study but the only information I was given was that a standard copper wire carrying either 12 V (.6 Amps) or 24 V (1.1 Amps) was wrapped around a solid cylindrical iron core (approx 40 turns) to create an electromagnet. There are two of these exact same devices a short distance apart (I've been starting at approx. 0.25 cm) attracting each other due to the parallel current in them. The only information I want is the Attractive Force (N) vs. Axial Distance (cm) of a few different iron core geometries, but I can't seem to get a sensible answer from the Electromagnetic force in the Global Data Display as I said before when I 'turned off' the iron core.
For some reason the iron has no influence... how do I fix this? Is there a better way to get more accurate axial force data?
Any and all suggestions welcome at this point. :)
Attachments:
10 Replies Last Post Jun 15, 2015, 11:19 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi
There is no reason why you should not see any effect by putting iron in your model, so obviously you must have got a boundary condition or a parameter wrong.
Now I would suggest that you use the symmetry and start in 2D axisymmetry with only 1/4 of the total system, you have a nice symmetric subject there, it's much quicker to start in 2D, and then run a full 3D at the end, just to check (with symmetry in models you easily forget a factor 2 or forget about antisymmetric modes i.e. structural analysis, so a final 3D check is a good safety rule and you learn quite alot on how to model too).
Be aware of a mistake one often do in 2D axisymmetry: it is to forget to multiply the integrands by 2*pi*r, the loop length, for postprocessing you have now a specific "tick box" to get volume integration, but you must explicitly add this factor in the Integration Coupling Variables.
I would try first with an "Azimuthal Induction Current, Vector Potential" 2D axisymmetric model
To analyse your symmetry part, you can use electric or magnetic isolation on your "lower" boundary conditions towards to other symmetric coil (similar to symmetry or antisymmetry in structural analysis).
Going one step further, you could use infinite elements for a external shell to get the flux lines running nicer and to reduce the surrounding air volume, hence reducing the number of mesh elements, three are a few examples in the documentation.
Magnetic isolation means that the current is flowing in the opposite direction in the symmetric part (repulsion forces) and the magnetic field runs parallel to the boundary, while electric isolation you will see the flux lines are perpendicular to the boundary (attractive forces).
If you run the model just with the coils in air you will notice that the Lorentz force are equal and have opposite signs depending on the magnetic isolation or electric isolation boundary condition, in accordance to the attraction or repulsion of two magnets. When you add your central soft iron the symmetry in attractive and repulsive forces disappears, just as you notice that the field lines are concentrated by the higher permittivity of the iron (I'm still not fully comfortable with this asymmetry in repulsion and attractive force with the iron, have to think more about that one myself).
To see the effect of the iron, you should turn on contour, arrow and streamline plots in the postprocessing.
Now coming back to your coil, I'm slightly puzzled there, you haven’t given all the details required to build the model, at least not from an engineering point of view:
You have 2 voltage and current cases both with some 20-21 Ohm resistance and 40 turns (?, both ?).
Assuming these are steady state values, with no motion nor back EMF (I could be wrong here), you have then either some 24 A-turn or 44 A-turn, but on the other hand you should have different wire gauges, hence different coil area and sizes.
At 1.1A I would expect about an AWG21 or M75 (diam 0.75mm) wire and for 0.6A AWG24 or M50 (diam 0.5mm). Certainly you could use smaller gauges, these values above are safe, but you will get quite some heating with smaller wires, hence changes in current and resistivity for your application.
Now this leads me to some 500 respectively 250 m wire length based on the resistance, for 40 turns that gives me a coil radius of 2 respectively 1 m, which is not what I believe to see on your figures ?
Furthermore the coil area, assuming a filling factor of 60% do neither not really fit the ratio.
You might have meant 400 turns but even then your 2 voltage - current cases should look quite different.
In any case try it out again, it's easier to debug in 2D, but in all cases run an analytical analysis to estimate the forces, just to be sure you have the correct order of magnitude, and even better measure and cross validate you model, its really easy to miss a little detail that gives you completely wrong answers
Good luck
Ivar
There is no reason why you should not see any effect by putting iron in your model, so obviously you must have got a boundary condition or a parameter wrong.
Now I would suggest that you use the symmetry and start in 2D axisymmetry with only 1/4 of the total system, you have a nice symmetric subject there, it's much quicker to start in 2D, and then run a full 3D at the end, just to check (with symmetry in models you easily forget a factor 2 or forget about antisymmetric modes i.e. structural analysis, so a final 3D check is a good safety rule and you learn quite alot on how to model too).
Be aware of a mistake one often do in 2D axisymmetry: it is to forget to multiply the integrands by 2*pi*r, the loop length, for postprocessing you have now a specific "tick box" to get volume integration, but you must explicitly add this factor in the Integration Coupling Variables.
I would try first with an "Azimuthal Induction Current, Vector Potential" 2D axisymmetric model
To analyse your symmetry part, you can use electric or magnetic isolation on your "lower" boundary conditions towards to other symmetric coil (similar to symmetry or antisymmetry in structural analysis).
Going one step further, you could use infinite elements for a external shell to get the flux lines running nicer and to reduce the surrounding air volume, hence reducing the number of mesh elements, three are a few examples in the documentation.
Magnetic isolation means that the current is flowing in the opposite direction in the symmetric part (repulsion forces) and the magnetic field runs parallel to the boundary, while electric isolation you will see the flux lines are perpendicular to the boundary (attractive forces).
If you run the model just with the coils in air you will notice that the Lorentz force are equal and have opposite signs depending on the magnetic isolation or electric isolation boundary condition, in accordance to the attraction or repulsion of two magnets. When you add your central soft iron the symmetry in attractive and repulsive forces disappears, just as you notice that the field lines are concentrated by the higher permittivity of the iron (I'm still not fully comfortable with this asymmetry in repulsion and attractive force with the iron, have to think more about that one myself).
To see the effect of the iron, you should turn on contour, arrow and streamline plots in the postprocessing.
Now coming back to your coil, I'm slightly puzzled there, you haven’t given all the details required to build the model, at least not from an engineering point of view:
You have 2 voltage and current cases both with some 20-21 Ohm resistance and 40 turns (?, both ?).
Assuming these are steady state values, with no motion nor back EMF (I could be wrong here), you have then either some 24 A-turn or 44 A-turn, but on the other hand you should have different wire gauges, hence different coil area and sizes.
At 1.1A I would expect about an AWG21 or M75 (diam 0.75mm) wire and for 0.6A AWG24 or M50 (diam 0.5mm). Certainly you could use smaller gauges, these values above are safe, but you will get quite some heating with smaller wires, hence changes in current and resistivity for your application.
Now this leads me to some 500 respectively 250 m wire length based on the resistance, for 40 turns that gives me a coil radius of 2 respectively 1 m, which is not what I believe to see on your figures ?
Furthermore the coil area, assuming a filling factor of 60% do neither not really fit the ratio.
You might have meant 400 turns but even then your 2 voltage - current cases should look quite different.
In any case try it out again, it's easier to debug in 2D, but in all cases run an analytical analysis to estimate the forces, just to be sure you have the correct order of magnitude, and even better measure and cross validate you model, its really easy to miss a little detail that gives you completely wrong answers
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hello again,
I hope you havn't given up ... your question got me to restudy the three AC/DC documents that have been quite updated since my last reading (V3.3). So now I'm also more confident why the repulsion and attraction forces were different, thinking about it, it's rather natural: iron as such is only attracted by the magnetic field (divergence) never repulsed, therefore the assymetry on the two forces. It comes from the B^2 term for reluctant force.
Now to extract the EM forces from COMSOL you have several methods, they are nicely described in the AC/DC documentation, apart that there is no "iron" in the examples. This adds a little suptility, you must select to either use the virtual displacement (more complex model set-up) but in my mind more precise, or use the Maxwell tensor approach, but then don't count the items twice, use the Lorentz force on your coil and Maxwell tensor items for the iron alone, which should add up to the same as (more or less to numerical and model approximations) to the Maxwell Tensor of iron+coil together.
Do not add up Lorentz forces + Maxwell tensor for the iron AND the coil, then you are too optimisitic (at least that's my conclusions).
The Maxwell tensor force for the coil agrees well with the Lorentz force without any iron, but is quite different when you add some iron along the coil. The model is no longer complete, I believe, normally from my EM books I see one should add up also the structural stress tensor, as the magnetic field (as to a less degree the electric field) add mechanical stress to the elastic material they are interacting with, by default matter is not elasticmin the AC/DC models, that requires to add another physics and to couple it in.
There is more fun and physics to study when you start to go into the details here, most of these effects were neglected, if not ignored before you could play with tools like COMSOL
So long
Ivar
I hope you havn't given up ... your question got me to restudy the three AC/DC documents that have been quite updated since my last reading (V3.3). So now I'm also more confident why the repulsion and attraction forces were different, thinking about it, it's rather natural: iron as such is only attracted by the magnetic field (divergence) never repulsed, therefore the assymetry on the two forces. It comes from the B^2 term for reluctant force.
Now to extract the EM forces from COMSOL you have several methods, they are nicely described in the AC/DC documentation, apart that there is no "iron" in the examples. This adds a little suptility, you must select to either use the virtual displacement (more complex model set-up) but in my mind more precise, or use the Maxwell tensor approach, but then don't count the items twice, use the Lorentz force on your coil and Maxwell tensor items for the iron alone, which should add up to the same as (more or less to numerical and model approximations) to the Maxwell Tensor of iron+coil together.
Do not add up Lorentz forces + Maxwell tensor for the iron AND the coil, then you are too optimisitic (at least that's my conclusions).
The Maxwell tensor force for the coil agrees well with the Lorentz force without any iron, but is quite different when you add some iron along the coil. The model is no longer complete, I believe, normally from my EM books I see one should add up also the structural stress tensor, as the magnetic field (as to a less degree the electric field) add mechanical stress to the elastic material they are interacting with, by default matter is not elasticmin the AC/DC models, that requires to add another physics and to couple it in.
There is more fun and physics to study when you start to go into the details here, most of these effects were neglected, if not ignored before you could play with tools like COMSOL
So long
Ivar
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi Andrew,
In my modeling work, I have placed a magnetic material in between two electromagnets and I would like to calculate the magnetic force between one electromagnet and the magnetic material. Therefore, will you please send me your modeling, so that I won't repeat the same mistake you had done in your work.
Regards,
Kiran
In my modeling work, I have placed a magnetic material in between two electromagnets and I would like to calculate the magnetic force between one electromagnet and the magnetic material. Therefore, will you please send me your modeling, so that I won't repeat the same mistake you had done in your work.
Regards,
Kiran
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi
I do ot have a model for that, but the main issue is, as I have learned:
if you use the Maxwell Tensor approach for forces, which is a simple and rather precise approach, then have some air around the material you want to study (even 100um or less, but take care for the meshing).
In other words do not have the coil volume (or the magnet) directly adjcent onto the iron yoke, leave a gap, anyhow you will have some "glue" in a true device.
My coil/iron test file is below
Good luck
Ivar
I do ot have a model for that, but the main issue is, as I have learned:
if you use the Maxwell Tensor approach for forces, which is a simple and rather precise approach, then have some air around the material you want to study (even 100um or less, but take care for the meshing).
In other words do not have the coil volume (or the magnet) directly adjcent onto the iron yoke, leave a gap, anyhow you will have some "glue" in a true device.
My coil/iron test file is below
Good luck
Ivar
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi,
Thanks a lot.
By the way, whether you model the iron and coil for electromagnet separately or you treated the coil as electromagnet and calculated the force between the electromagnet and iron.
Actually, my case is little bit complicated. In my model, a foam will be placed in between the iron and the electromagnet. In this configuration, I will calculate the force between the electromagnet and iron.
If you have any idea on this, please pass it to me which will be a great help.
Thanks Again,
Kiran
Thanks a lot.
By the way, whether you model the iron and coil for electromagnet separately or you treated the coil as electromagnet and calculated the force between the electromagnet and iron.
Actually, my case is little bit complicated. In my model, a foam will be placed in between the iron and the electromagnet. In this configuration, I will calculate the force between the electromagnet and iron.
If you have any idea on this, please pass it to me which will be a great help.
Thanks Again,
Kiran
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi
Your foam is probably similar enough to air so this should then not influence much the results of the force calculations even with the Maxwell stress tensor, run your model with air and then with the foam and check the differences
you have always the "virtual work" approach, the model set-up is somewhat more complex, see the examle:
v3.5a "acdcmodelib.pdf" page 8, "Electromagnetic Forces on Parallel Current-Carrying Wires"
and read through “Electromagnetic Forces” on page 133 in the AC/DC Module User’s Guide,
good luck
Ivar
Your foam is probably similar enough to air so this should then not influence much the results of the force calculations even with the Maxwell stress tensor, run your model with air and then with the foam and check the differences
you have always the "virtual work" approach, the model set-up is somewhat more complex, see the examle:
v3.5a "acdcmodelib.pdf" page 8, "Electromagnetic Forces on Parallel Current-Carrying Wires"
and read through “Electromagnetic Forces” on page 133 in the AC/DC Module User’s Guide,
good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi,
First of all thanks a lot for your help.
I have one more question. An electromagnet consists of a coil wounded on a feromagnetic material. In your geometric model, both the solenoid and the coil have same dimension, which I couldn't undersatand.
One more thing, in the geometric model you have considered the coil as electromagnet i.e., only the coil will produce an electromagnetic field, which is not true as fas my knowledge is concerned. Selonoid which is made of a ferromagnetic material will also produce an electromagnetic field in addition to the coil.
I may be wrong. However, please rectify my doubt.
Thanks again.
Kiran
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi
I do not understand you here.
The model has a small Copper"square" coil" in which I send some current making a given magnetic field in the volume (and in the coil). The iron (ferromagnetic) rectangular part only concetrates the field due to its high magnetic permeability (I used 4000 just like tht) the *Param is there to first calculate the electromagnetic field without any iron (Param=0), then to add the iron permeativity gently just to illustrate how an increasing ferromagtic material acts on the magnetic filed lines
Ivar
I do not understand you here.
The model has a small Copper"square" coil" in which I send some current making a given magnetic field in the volume (and in the coil). The iron (ferromagnetic) rectangular part only concetrates the field due to its high magnetic permeability (I used 4000 just like tht) the *Param is there to first calculate the electromagnetic field without any iron (Param=0), then to add the iron permeativity gently just to illustrate how an increasing ferromagtic material acts on the magnetic filed lines
Ivar
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi everyone,
I'm new to COMSOL and have a problem with electrostatic force calculation. The initial problem concerned the force between a charged metal point and a polarized dielectric material (through the gradient of the E field). The results were non-realistic so I resorted to testing the force calculations using a simple model of two charged sphere's that repel one another. Again I got non-realistic results!!! This sounds very simple to do and I'm sure I've missed something somewhere but I can't seem to find my error.
Here's the problem.
In 3D "emes" module create two small charged sphere's separated by some distance "d". Then enclose the sphere's by a much larger box filled with air. Calculate the force between them.
The force on the spheres should follow coulombs law F=k*q^2/d^2 and obviously it's direction should be parallel to the line joining the two charges
This problem seems so easy and yet I can't find where I've gone wrong.
Cheers
Glen
I'm new to COMSOL and have a problem with electrostatic force calculation. The initial problem concerned the force between a charged metal point and a polarized dielectric material (through the gradient of the E field). The results were non-realistic so I resorted to testing the force calculations using a simple model of two charged sphere's that repel one another. Again I got non-realistic results!!! This sounds very simple to do and I'm sure I've missed something somewhere but I can't seem to find my error.
Here's the problem.
In 3D "emes" module create two small charged sphere's separated by some distance "d". Then enclose the sphere's by a much larger box filled with air. Calculate the force between them.
The force on the spheres should follow coulombs law F=k*q^2/d^2 and obviously it's direction should be parallel to the line joining the two charges
This problem seems so easy and yet I can't find where I've gone wrong.
Cheers
Glen
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hi Ivar,
I am facing similar problem in calculating the maxwell tensile force in comsol. Its formula from literature is F=B^2*A/2*muo. The details of my geometry are as follows.
I want to calculate the magnetic force between one of my pole of electromagnet (Winding is done on a magnetic material) and a cylinder (made from magnetic material). There is a foam between surface of electromagnet and inner surface of cylinder.
I found magnetic flux at the surface of foam, area=2*pi*r*l and calculate its magnetic force. But I want to find it in Comsol. I also verified this force experimentally through this formula F=B^2*A/2*muo, but couldn't do so through comsol.
Your help in this regard would be highly appreciated. Thanks
I am facing similar problem in calculating the maxwell tensile force in comsol. Its formula from literature is F=B^2*A/2*muo. The details of my geometry are as follows.
I want to calculate the magnetic force between one of my pole of electromagnet (Winding is done on a magnetic material) and a cylinder (made from magnetic material). There is a foam between surface of electromagnet and inner surface of cylinder.
I found magnetic flux at the surface of foam, area=2*pi*r*l and calculate its magnetic force. But I want to find it in Comsol. I also verified this force experimentally through this formula F=B^2*A/2*muo, but couldn't do so through comsol.
Your help in this regard would be highly appreciated. Thanks