I don't the formula for circulating the energy loss but you can look it up and see what parameters are required. The B field will be one of them and you can plot any required parameters in the result section. You can define the formula in Comsol to calculate the loss for each simulation.
Is it better to use infinite element domains all the time vs just a box or a sphere for magnetic simulations? Is one faster to compute than the other? Also is there a difference between a box and a sphere as the outer boundary of the simulation in general?
We need to give enough room for the magnetic field lines to spread in space. If you do this by creating a very large box/sphere then that's going to be more computationally intensive than using infinite element domain on the boundary of a small region. So in this regard, infinite element domain is always better to use. If you use infinite element domain, you can use either a rectangular or spherical region but choose the appropriate coordinate system.
@@JafScience Thanks. That makes sense. So if the modeled geometry is rectangular, it's actually better to use a rectangular boundary, because there would be less volume to mesh between the model and the outer surface, right? Same with a spherical model and spherical boundary. One more question. If I need to simulate hysteresis effects vs frequency, which model should I use? Jiles-Atherton is built into COMSOL if I understand correctly, but it doesn't consider frequency, right? Maybe I need to find a model that does and manually enter it into the simulation somehow?
@@batsali99 For your first question, yes, just match the type of coordinate system with the type of air domain region. For question 2, the frequency domain solver gives you the steady-state harmonic response while hysteresis is a time-dependent phenomenon. There may be some tricks published in the literature to help you work around this problem but Comsol doesn't have the capability to model hysteresis in the frequency domain.
@@JafScience I want to study the spectrum of distortion products due to the material properties of a transformer core for an input excitation that is also fairly harmonically rich. The known parameters would be the B-H curve of the core, probably a fit of Jiles-Atherton parameters to measurements (or some other model?). Is that not possible with Comsol? If the input excitation is defined as a sum of a few sines in time it should work, no? I've been reading a bit more and things like 'magnetic viscosity' are popping up. Maybe this is a much more difficult problem than I thought. If it's not possible with Comsol, do you know of any other software that would be useful for this application? Thanks for taking the time to answer!
@@batsali99 You could do several frequency domain studies at different frequencies and then use inverse FFT to transform your results to the time domain. But in each Study, the Jiles-Atherton model can't be used. The limitation is mainly mathematical and is not due to Comsol being inadequate so I don't any other software that would be able to do what you want.
If you mean the H field due to magnetization M then that's already included. The H field you see in Comsol is the total (sum of external and demagnetizing fields). I hope this helps.
Dear Jaf, Your tutorials are impressive and very helpful. Could you do a tutorial on how to simulate magnetic hyperthermia of nanoparticles via comsol? Although I know which physics controls the process, I have a hard time building the model that converges.
I can show an example sometime in the future. If you want to create a model now, you can either use the helix geometry (set axial spacing to zero to create the pancake coil) or you can create concentric circles which represent the coil loops. In physics just apply one edge current node to all the loops (or 2 nodes for bifilar coil).
Circuit voltage fixes the voltage and calculates current using Ohm's law. Circuit current fixes the current and calculates the voltage accordingly. In high frequency AC applications, the two options can give different results because the effective resistance of the current carrier changes with frequency (skin effect). In such cases, Circuit current preserves total electrical energy, Circuit voltage does not (energy is lost).
First of all thank you for such a great presentation. Secondly, if the material does not use the JA model but uses the BH curve, can the hysteresis loop be drawn?
No, a BH curve does not capture the hysteretic losses. Even if you import such a BH curve from experiment, it would be limited to the particular cycle which you imported.
@@JafScience Thanks I got it. You mean I still have to use the JA model, but there are several parameters in the model that need to be set, which I can't find from the material manual used
Yes, the JA parameters are obtained experimentally. One possible solution for you is to pick a hysteretic BH curve and find the parameters which can reproduce that curve. You'll need to use an appropriate input magnetic field.
Hi Jaf-Science! First of all, I have to say that this is an incredible tutorial. Thank you so much for sharing all these invaluable technical information, much appreciated. Second, I would like to know if it is possible to simulate the heat generation through magnetic hysteresis by coupling magnetic fields physics with the heat transfer in solids physics? Regards...
Yes you can couple those physics interfaces. You need to define a variable for the hysteretic losses in the variables section. This variable can be used as a heat source in the heat transfer physics.
Dear Jaf, your tutorials are really helpful. But i have a problem with this one. I repeat everything you did and have an error "Repeated error test failures. May have reached a singularity.". Do you know how to fix this? Comsol 6.1
The video was made using v5.6. Can you run the simulation using constant permeability? See if that works. Then go back to the nonlinear BH curve. Alternatively try to build the model again from scratch. Sometimes something is missed and it becomes difficult to detect.
Same error message shown as I was using 6.1. Anyway, I changed the electrical conductivity in the Jiles-Atherton Hysteretic Material, "0 → 0.1"...it worked...not showing any error.
Thanks for this tutorial, it was great. I have a question about thermal effect on the permanent magnet, I couldnt find a way to add temprature to see how does it effect on the magnetic flux density. If you can help me i will be very thankful.
The magnetic field produced by the magnet decreases with increasing temperature. You need to import such data into a function in Comsol because it cannot be computed in Comsol. You can look for temperature curves from a manufacturer. kjmagnetics is one supplier and their website is rich in information about magnets.
@@JafScience thank you, But i couldnt find sth that can helps me through that website. Coukd you please explain a little more? If there s no parametr related to tempature in magnetic field physics node, how can i write a function ?
If you're working on a multilhysics problem then you'll need to insert the heat transfer interface. T will be the temperature variable. The graphs on the website shows different curves for each temperature.
@@JafScience Thank you. I have another question, about finding center mass of a geometry. There should be a mass propertise subnode in definitions node, but in comsol 5.5 i couldnt find it!!! Can you help me Thank you
I'm not aware of a direct feature to give you centre of mass but you can find it using the integration operator. Insert integration under definition and under variables define total mass: M=density* intop1(1). Then define the following: Sx=density*intop1(x). Then the x coordinate of Centre of mass is Sx/M. Do the same steps for y and z coordinates.
Yes, if you create a spherical air space then you'll select spherical type for infinite element domain. The geometry of the ferromagnet can be anything.
@@JafScience while applying another material on the ferromagnetic cylinder , and selecting jiles atherton model in property group, even when changing the values of jiles atherton parameters , the hysteresis loop remains the same. Only the saturation magnetisation is changing. Is it normal ? 😕🙁
You mean the shape of the hysteresis loop remains the same when you change the Jiles Atherton parameters? It is supposed to change. For instance, if you increase reversibility then the loop will become narrower. Also make sure the software is reading the parameters you have defined. This is either in the physics or in the material section (there should be a green check mark next to the property if Comsol is using it).
For my question, imagine we are at the point (-Hc,0). What will be the new path if I start increasing H instead of letting the curve enter the 3rd quadrant? From your simulation, it seems like the curve would always follow an almost horizontal path before becoming steep everytime you change directions... in my head the curve that has its direction changed at (-Hc,0) should just be a translation of the curve that starts at (0,0) since both have no magnetization. Can you clarify my thoughts???
Interesting insight! Just to clarify, are you expecting the curve to go up from (-Hc, 0) but have the same shape as the curve that starts at (0, 0)? I would expect an almost horizontal line going back to (0, 0) since Hc is the coercive field, defined as the field strength required to demagnetize a ferromagnetic material.
@@JafScience Yes, that is what I'm expecting. That is because the physical situation is the same as when the curve starts from (0,0), in other words, there is no net magnetic flux within the core. So... reversing H from that point should result in the same curve it would if we kept going negative (which is actually the same shape as the one that starts at (0,0) but translated and flipped) from the core's point of view it should provide the same "resistance" independent of increasing towards north or south pole. But again, that's just a thought I had. From the results shown, that's probably wrong, but again, I don't understand why
If the curve starts going up at (-Hc,0), that would imply that the material is still magnetised. But theory says that the material is demagnetised once you apply a coercive field. A demagntised material will have a BH curve starting at the origin when an external magnetic field is applied.
@@JafScience I didn't get why the curve going up implies that the material is magnetized... also, you said that when a coercive field is applied, the material is demagnetized, and then said that a demagnetized material has its curve starting at the origin. What I understand from those statements together is that the material behaves the same at (-Hc,0) and (0,0), which doesn't seem to be the case according to the simulation...
The purpose of this video is just to show the hysteresis effect and how to model it. Practical importance depends on the application you want to model.
This is a perfect video that anyone can follow completely. Thanks to your simple and clear explanation, I was able to learn it quickly.
This is very helpful for real
Thank you very much for the explanation. Would you please explain how we can extract hysteresis loss from the model.
I don't the formula for circulating the energy loss but you can look it up and see what parameters are required. The B field will be one of them and you can plot any required parameters in the result section. You can define the formula in Comsol to calculate the loss for each simulation.
Thanks a lot! great job as usual.
Thank you for such nice explanation.
Is it better to use infinite element domains all the time vs just a box or a sphere for magnetic simulations? Is one faster to compute than the other? Also is there a difference between a box and a sphere as the outer boundary of the simulation in general?
We need to give enough room for the magnetic field lines to spread in space. If you do this by creating a very large box/sphere then that's going to be more computationally intensive than using infinite element domain on the boundary of a small region. So in this regard, infinite element domain is always better to use.
If you use infinite element domain, you can use either a rectangular or spherical region but choose the appropriate coordinate system.
@@JafScience Thanks. That makes sense. So if the modeled geometry is rectangular, it's actually better to use a rectangular boundary, because there would be less volume to mesh between the model and the outer surface, right? Same with a spherical model and spherical boundary.
One more question. If I need to simulate hysteresis effects vs frequency, which model should I use? Jiles-Atherton is built into COMSOL if I understand correctly, but it doesn't consider frequency, right? Maybe I need to find a model that does and manually enter it into the simulation somehow?
@@batsali99 For your first question, yes, just match the type of coordinate system with the type of air domain region.
For question 2, the frequency domain solver gives you the steady-state harmonic response while hysteresis is a time-dependent phenomenon. There may be some tricks published in the literature to help you work around this problem but Comsol doesn't have the capability to model hysteresis in the frequency domain.
@@JafScience I want to study the spectrum of distortion products due to the material properties of a transformer core for an input excitation that is also fairly harmonically rich. The known parameters would be the B-H curve of the core, probably a fit of Jiles-Atherton parameters to measurements (or some other model?). Is that not possible with Comsol? If the input excitation is defined as a sum of a few sines in time it should work, no? I've been reading a bit more and things like 'magnetic viscosity' are popping up. Maybe this is a much more difficult problem than I thought. If it's not possible with Comsol, do you know of any other software that would be useful for this application? Thanks for taking the time to answer!
@@batsali99 You could do several frequency domain studies at different frequencies and then use inverse FFT to transform your results to the time domain. But in each Study, the Jiles-Atherton model can't be used. The limitation is mainly mathematical and is not due to Comsol being inadequate so I don't any other software that would be able to do what you want.
Thanks for your great explanation. Could you please explain how you would include the demagnetization fields in the model?
If you mean the H field due to magnetization M then that's already included. The H field you see in Comsol is the total (sum of external and demagnetizing fields). I hope this helps.
Hi, very useful video, thanks. Do you have an example of how to evaluate the core loss using the jiles-atherton model?
I haven't made such an example.
That is great work.I have question can be simulate trilayer for get decoupling or bilayer to get exchange bias?
Very useful thank you
Dear Jaf,
Your tutorials are impressive and very helpful.
Could you do a tutorial on how to simulate magnetic hyperthermia of nanoparticles via comsol? Although I know which physics controls the process, I have a hard time building the model that converges.
I don't have knowledge of modelling nanoparticles, sorry.
Thanks a lot for your great job.
Could you please give an example on pancake coil (single coil or bifilar).
I can show an example sometime in the future. If you want to create a model now, you can either use the helix geometry (set axial spacing to zero to create the pancake coil) or you can create concentric circles which represent the coil loops. In physics just apply one edge current node to all the loops (or 2 nodes for bifilar coil).
@@JafScience thank you. i am going to try.
Could you please tell me the difference between circuit [voltage)and circuit [current] in coil excitation
Circuit voltage fixes the voltage and calculates current using Ohm's law. Circuit current fixes the current and calculates the voltage accordingly. In high frequency AC applications, the two options can give different results because the effective resistance of the current carrier changes with frequency (skin effect). In such cases, Circuit current preserves total electrical energy, Circuit voltage does not (energy is lost).
Thank you sir
First of all thank you for such a great presentation. Secondly, if the material does not use the JA model but uses the BH curve, can the hysteresis loop be drawn?
No, a BH curve does not capture the hysteretic losses. Even if you import such a BH curve from experiment, it would be limited to the particular cycle which you imported.
@@JafScience Thanks I got it. You mean I still have to use the JA model, but there are several parameters in the model that need to be set, which I can't find from the material manual used
Yes, the JA parameters are obtained experimentally. One possible solution for you is to pick a hysteretic BH curve and find the parameters which can reproduce that curve. You'll need to use an appropriate input magnetic field.
@@JafScience Thanks, I see. Thanks again for answering my questions.
Hi Jaf-Science! First of all, I have to say that this is an incredible tutorial. Thank you so much for sharing all these invaluable technical information, much appreciated. Second, I would like to know if it is possible to simulate the heat generation through magnetic hysteresis by coupling magnetic fields physics with the heat transfer in solids physics? Regards...
Yes you can couple those physics interfaces. You need to define a variable for the hysteretic losses in the variables section. This variable can be used as a heat source in the heat transfer physics.
@@JafScience How would you define this variable?
@@kingmike134 Check my transformer tutorial to see how to define variables.
Can you show for magnetostriction
Dear Jaf, your tutorials are really helpful. But i have a problem with this one. I repeat everything you did and have an error "Repeated error test failures. May have reached a singularity.". Do you know how to fix this? Comsol 6.1
Does your solver have the same settings as in the video? Not sure if version 6.1 has different settings by default.
Mb. Did step by step as you. Which version you have on this video? Really need this project to work. It can help me pass the exam
The video was made using v5.6. Can you run the simulation using constant permeability? See if that works. Then go back to the nonlinear BH curve. Alternatively try to build the model again from scratch. Sometimes something is missed and it becomes difficult to detect.
Same error message shown as I was using 6.1. Anyway, I changed the electrical conductivity in the Jiles-Atherton Hysteretic Material, "0 → 0.1"...it worked...not showing any error.
Thanks for this tutorial, it was great.
I have a question about thermal effect on the permanent magnet, I couldnt find a way to add temprature to see how does it effect on the magnetic flux density. If you can help me i will be very thankful.
The magnetic field produced by the magnet decreases with increasing temperature. You need to import such data into a function in Comsol because it cannot be computed in Comsol. You can look for temperature curves from a manufacturer. kjmagnetics is one supplier and their website is rich in information about magnets.
@@JafScience thank you,
But i couldnt find sth that can helps me through that website.
Coukd you please explain a little more?
If there s no parametr related to tempature in magnetic field physics node, how can i write a function ?
If you're working on a multilhysics problem then you'll need to insert the heat transfer interface. T will be the temperature variable.
The graphs on the website shows different curves for each temperature.
@@JafScience
Thank you.
I have another question, about finding center mass of a geometry.
There should be a mass propertise subnode in definitions node, but in comsol 5.5 i couldnt find it!!!
Can you help me
Thank you
I'm not aware of a direct feature to give you centre of mass but you can find it using the integration operator. Insert integration under definition and under variables define total mass: M=density* intop1(1). Then define the following: Sx=density*intop1(x). Then the x coordinate of Centre of mass is Sx/M. Do the same steps for y and z coordinates.
Is the model file available somewhere, or one needs to recreate it from scratch?
No it's not uploaded. You can create it from scratch.
Is this can be done in 3D model as well ?
By choosing the spherical geometry type instead of cylindrical.
Yes, if you create a spherical air space then you'll select spherical type for infinite element domain. The geometry of the ferromagnet can be anything.
@@JafScience thanks !!! 👌🏾
@@JafScience while applying another material on the ferromagnetic cylinder , and selecting jiles atherton model in property group, even when changing the values of jiles atherton parameters , the hysteresis loop remains the same.
Only the saturation magnetisation is changing.
Is it normal ? 😕🙁
You mean the shape of the hysteresis loop remains the same when you change the Jiles Atherton parameters? It is supposed to change. For instance, if you increase reversibility then the loop will become narrower. Also make sure the software is reading the parameters you have defined. This is either in the physics or in the material section (there should be a green check mark next to the property if Comsol is using it).
@@JafScience yes i mean the shape of hysteresis loop doesn’t change.
Still not working 🙁
For my question, imagine we are at the point (-Hc,0). What will be the new path if I start increasing H instead of letting the curve enter the 3rd quadrant? From your simulation, it seems like the curve would always follow an almost horizontal path before becoming steep everytime you change directions... in my head the curve that has its direction changed at (-Hc,0) should just be a translation of the curve that starts at (0,0) since both have no magnetization. Can you clarify my thoughts???
Interesting insight! Just to clarify, are you expecting the curve to go up from (-Hc, 0) but have the same shape as the curve that starts at (0, 0)? I would expect an almost horizontal line going back to (0, 0) since Hc is the coercive field, defined as the field strength required to demagnetize a ferromagnetic material.
@@JafScience Yes, that is what I'm expecting. That is because the physical situation is the same as when the curve starts from (0,0), in other words, there is no net magnetic flux within the core. So... reversing H from that point should result in the same curve it would if we kept going negative (which is actually the same shape as the one that starts at (0,0) but translated and flipped) from the core's point of view it should provide the same "resistance" independent of increasing towards north or south pole. But again, that's just a thought I had. From the results shown, that's probably wrong, but again, I don't understand why
If the curve starts going up at (-Hc,0), that would imply that the material is still magnetised. But theory says that the material is demagnetised once you apply a coercive field. A demagntised material will have a BH curve starting at the origin when an external magnetic field is applied.
@@JafScience I didn't get why the curve going up implies that the material is magnetized... also, you said that when a coercive field is applied, the material is demagnetized, and then said that a demagnetized material has its curve starting at the origin. What I understand from those statements together is that the material behaves the same at (-Hc,0) and (0,0), which doesn't seem to be the case according to the simulation...
@@JafScience btw, don't get me wrong, I'm just trying to learn this hysteresis concept fully, I have a really tough time trying XD
I couldn't find jearls arthen model in my cosmol. How can I get it
Versions older than 5.2 did not have Jiles-Atherton model. e
@@JafScience I have 5.1 version. Now instead of jearls arthon model which alternative model can I use for modeling hysteresis
It's not possible without defining a mathematical model for hysteresis. This is a difficult route and there isn't an easy alternative.
Why we need to draw different cycles of hysteresis, I mean the importance of minor loops.
The purpose of this video is just to show the hysteresis effect and how to model it. Practical importance depends on the application you want to model.
@@JafScience ok sir
@@JafScience thanks 😊
thx