Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.
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.
Coupled magnetic and solid-mechanics simulation
Posted May 9, 2012, 7:52 a.m. EDT Low-Frequency Electromagnetics, Modeling Tools & Definitions, Parameters, Variables, & Functions, Structural Mechanics Version 4.4 17 Replies
Please login with a confirmed email address before reporting spam
I have a question about coupling a magnetic study to a solid mechanics study.
I attach to this email a simple model to explain my question. In the model you can find a permanent magnet with a round soft-iron path to guide the magnetic flux, and an air gap (and an air volume surrounding it). I added electromagnetic physics and solid mechanics physics, and I would like to perform a fully coupled study. That is, the magnet generates magnetic flux, which in the air-gap generates a force. That force deforms the soft-iron, therefore reducing the air-gap and increasing the flux. Thus, there is 2 couplings: the magnetic force deforms the solid object, and the deformations change the magnetic path.
What I see when running the simulation (study 3) is that I can indeed deform the structure thanks to the magnetic force. Nevertheless, I don’t think the magnetic flux is being recalculated after the deformation. To verify this I added a boundary load to the solid mechanics physics which pushes the air-gap closed. When performing the simulation, I do see a difference in the deformation, but I see no difference in the flux (you can check it with the surface average node in the derived values). The same happens if I chance the young modulus of the soft –iron, even to un-realistic values: no matter the final size of the air-gap, the flux doesn’t change. I tried a fully coupled study (instead of segregated) and get the same result.
Do you know what am I doing wrong?
A second question is that when I run the simulation with a force of 0 or 200N (you can change it in the parameters) it runs fast and easy, but if I try a parametric study with those 2 values, it gets stuck. You can see it in the study 2.
I attach to this email a simple model to explain my question. In the model you can find a permanent magnet with a round soft-iron path to guide the magnetic flux, and an air gap (and an air volume surrounding it). I added electromagnetic physics and solid mechanics physics, and I would like to perform a fully coupled study. That is, the magnet generates magnetic flux, which in the air-gap generates a force. That force deforms the soft-iron, therefore reducing the air-gap and increasing the flux. Thus, there is 2 couplings: the magnetic force deforms the solid object, and the deformations change the magnetic path.
What I see when running the simulation (study 3) is that I can indeed deform the structure thanks to the magnetic force. Nevertheless, I don’t think the magnetic flux is being recalculated after the deformation. To verify this I added a boundary load to the solid mechanics physics which pushes the air-gap closed. When performing the simulation, I do see a difference in the deformation, but I see no difference in the flux (you can check it with the surface average node in the derived values). The same happens if I chance the young modulus of the soft –iron, even to un-realistic values: no matter the final size of the air-gap, the flux doesn’t change. I tried a fully coupled study (instead of segregated) and get the same result.
Do you know what am I doing wrong?
A second question is that when I run the simulation with a force of 0 or 200N (you can change it in the parameters) it runs fast and easy, but if I try a parametric study with those 2 values, it gets stuck. You can see it in the study 2.
Attachments:
17 Replies Last Post Jan 23, 2016, 5:21 p.m. EST
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
you are right that you do not couple, since the deformation of the solid does NOT deform the mesh such that the B field might be adapted. Even if you solve each together in a combined way. You need to use ALE to get that going.
Use a "continuation sweep and not a specific parametric sweep, if you have only BC values to change (a continuation sweep uses the last values from previous simulation as starting values for the next, ideal for an increasing force case), while a parametric sweep node updates the geoemtry, the mesh and restarts each time with original initial conditions "0"
check the solver tabs at the end of the window (its by default closed and must be opened)
--
Good luck
Ivar
you are right that you do not couple, since the deformation of the solid does NOT deform the mesh such that the B field might be adapted. Even if you solve each together in a combined way. You need to use ALE to get that going.
Use a "continuation sweep and not a specific parametric sweep, if you have only BC values to change (a continuation sweep uses the last values from previous simulation as starting values for the next, ideal for an increasing force case), while a parametric sweep node updates the geoemtry, the mesh and restarts each time with original initial conditions "0"
check the solver tabs at the end of the window (its by default closed and must be opened)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I'm doing a magnetic-structural coupling model. Thanks for your early comment. I've built a model with three physics,
1. Magnetic fiels no current model
2. Solid-mechanics model
3. Moving mesh model
I wonder if moving mesh is the ALE you mentioned before.
I can run either magnetic model or solid mechanics model. Yet when I tried to couple them together, the coupling model was not right. I just used the output force of magnetic model as the input load of solid mechanics model. For example, I inserted 'mfnc.Forcex_rod' into the body force of solid mechanics model as an input load. 'mfnc.Forcex_rod' is the force I got from the magnetic model. I dould see the results before coupling. I thought the models were not coupled at all.
Another issue was about 'continuation sweep' setting. When chose the setting and tried to set sweep type as specific combination. Nothing came out under the 'continuation parameter'. I thought I might miss certain steps before this. Yet I'm not sure it's the meshing problem or other issues.
Thanks a lot.
I'm doing a magnetic-structural coupling model. Thanks for your early comment. I've built a model with three physics,
1. Magnetic fiels no current model
2. Solid-mechanics model
3. Moving mesh model
I wonder if moving mesh is the ALE you mentioned before.
I can run either magnetic model or solid mechanics model. Yet when I tried to couple them together, the coupling model was not right. I just used the output force of magnetic model as the input load of solid mechanics model. For example, I inserted 'mfnc.Forcex_rod' into the body force of solid mechanics model as an input load. 'mfnc.Forcex_rod' is the force I got from the magnetic model. I dould see the results before coupling. I thought the models were not coupled at all.
Another issue was about 'continuation sweep' setting. When chose the setting and tried to set sweep type as specific combination. Nothing came out under the 'continuation parameter'. I thought I might miss certain steps before this. Yet I'm not sure it's the meshing problem or other issues.
Thanks a lot.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Dan,
Unfortunately I don't have the model at hand anymore. Anyway, for what I remember indeed ALE is the moving mesh, and you have to couple that one to the displacements of the mechanics model, and the forces from magnetic to the mechanic model. That is, the magnetic runs and generates the forces, the mechanic runs with those forces and generate displacements, and then the ALE runs with those displacements and generates the new mesh, and ready for the next cycle.
I hope this helps!
Unfortunately I don't have the model at hand anymore. Anyway, for what I remember indeed ALE is the moving mesh, and you have to couple that one to the displacements of the mechanics model, and the forces from magnetic to the mechanic model. That is, the magnetic runs and generates the forces, the mechanic runs with those forces and generate displacements, and then the ALE runs with those displacements and generates the new mesh, and ready for the next cycle.
I hope this helps!
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Pablo,
Thanks very much for your advice. Would you mind checing my model?
Thanks very much for your advice. Would you mind checing my model?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Would you mind taking a loot at the model?
Thanks.:)
Thanks.:)
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Unfortunately I don't have comsol installed in this machine anymore.
Best luck and regards,
Pablo
Best luck and regards,
Pablo
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dan,
There are multiple issues in your “permanent_magnet.mph” setup.
Some (not all) issues are listed below.
Solid Mechanics:
- Air domain should not be included
- Body load should be applied to the solid part which is moving (not to all domains)
- Replace Fixed Constraint by appropriate constraints along symmetry planes
- Prescribed Displacement node is not needed since movement of the solid body is governed by Body Load
Moving Mesh:
- Free Deformation should include only domain where is deformed, i.e. air domain
- Symmetry constraints are needed for two symmetry planes. These constraints are imposed by Prescribed Mesh Displacement Nodes
-There is no coupling between Mowing Mesh and Solid Mechanics nodes
Attached is working mph file of your model for the case when magnet is fixed and rod is moving. This is not an attempt to solve your specific problem but to show general setup steps for Magnetic/Solid/Moving Mesh coupling.
Note, that you can replace Solid Mechanics node by ODE and DAE Interface if you are not interested in deformation of moving solid – see example “Magnet Falling through Copper Tube” from Comsol Model Library.
Regards,
Sergei
There are multiple issues in your “permanent_magnet.mph” setup.
Some (not all) issues are listed below.
Solid Mechanics:
- Air domain should not be included
- Body load should be applied to the solid part which is moving (not to all domains)
- Replace Fixed Constraint by appropriate constraints along symmetry planes
- Prescribed Displacement node is not needed since movement of the solid body is governed by Body Load
Moving Mesh:
- Free Deformation should include only domain where is deformed, i.e. air domain
- Symmetry constraints are needed for two symmetry planes. These constraints are imposed by Prescribed Mesh Displacement Nodes
-There is no coupling between Mowing Mesh and Solid Mechanics nodes
Attached is working mph file of your model for the case when magnet is fixed and rod is moving. This is not an attempt to solve your specific problem but to show general setup steps for Magnetic/Solid/Moving Mesh coupling.
Note, that you can replace Solid Mechanics node by ODE and DAE Interface if you are not interested in deformation of moving solid – see example “Magnet Falling through Copper Tube” from Comsol Model Library.
Regards,
Sergei
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
another issue I have been told: set the moveing mesh ALE physics first, before the structural, as node order / precedence has also some importance. Solid and its frames, and ALE do interfer.
As clearly stated in the reply above, we need to analyse our models first, in detail: identify clearly the links, see which dependent variables are solved where and identify how to link them.
Multiphyiscs, is somewhat more complex than what one could first think, but once used to it, it becomes real fun ;)
--
Good luck
Ivar
another issue I have been told: set the moveing mesh ALE physics first, before the structural, as node order / precedence has also some importance. Solid and its frames, and ALE do interfer.
As clearly stated in the reply above, we need to analyse our models first, in detail: identify clearly the links, see which dependent variables are solved where and identify how to link them.
Multiphyiscs, is somewhat more complex than what one could first think, but once used to it, it becomes real fun ;)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Sergei,
Thanks so much for your suggestions and your guidance on my model.
I got one more question about the difference between "segregated step" and 'fully coupled' command. I noticed in a lot of coupling analysis. Fully coupled was used. I am thinking segregated step is we manully divide and set the running order of different physics. Is it right?
Thanks very much.
Thanks so much for your suggestions and your guidance on my model.
I got one more question about the difference between "segregated step" and 'fully coupled' command. I noticed in a lot of coupling analysis. Fully coupled was used. I am thinking segregated step is we manully divide and set the running order of different physics. Is it right?
Thanks very much.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Ivar,
Thanks so much.
I used to set ALE after all the physics, thinking mesh will only be changed after the real deformation from the solid mechanics. Yet I was totally doing the wrong way.
One more thing I want to ask is about 'continuation sweep'. Since I've used ALE, I'm thinking there's no need to use continuation sweep. Is it right?
Thanks very much.:)
Thanks so much.
I used to set ALE after all the physics, thinking mesh will only be changed after the real deformation from the solid mechanics. Yet I was totally doing the wrong way.
One more thing I want to ask is about 'continuation sweep'. Since I've used ALE, I'm thinking there's no need to use continuation sweep. Is it right?
Thanks very much.:)
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
a continuation sweep is useful if you want to step through a solution controlling a value by a parameter to scan it, but for your case I do not see what tat will bring. Often when you impose a displacement, it's worth to go by steps, also with ALE
--
Good luck
Ivar
a continuation sweep is useful if you want to step through a solution controlling a value by a parameter to scan it, but for your case I do not see what tat will bring. Often when you impose a displacement, it's worth to go by steps, also with ALE
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
another issue I have been told: set the moveing mesh ALE physics first, before the structural, as node order / precedence has also some importance. Solid and its frames, and ALE do interfer.
As clearly stated in the reply above, we need to analyse our models first, in detail: identify clearly the links, see which dependent variables are solved where and identify how to link them.
Multiphyiscs, is somewhat more complex than what one could first think, but once used to it, it becomes real fun ;)
--
Good luck
Ivar
Hi,
When I set my moving mesh ALE node before the solid mechanics node, all the domains I had chosen in the moving mesh become not applicable. Please see attached screenshot.
Thanks,
Jeff
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
that is exactly it, Solid has its own frame system so this overides ALE, ALE is for air (which is no solid) domain that is deformed by the solid deformation (check the model library examples)
--
Good luck
Ivar
that is exactly it, Solid has its own frame system so this overides ALE, ALE is for air (which is no solid) domain that is deformed by the solid deformation (check the model library examples)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Can someone give me a successful example about coupling magnetic and soild-mechanics, I am trying this, but have no idea now.
Thank you!
Thank you!
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Dr. Sergei,
I have attached a model. In this model I have two magnets which can be the same polarity or opposite.
The magnets are in tip of the rods, and I want to see the rods deform when the magnets have same or opposite polarities.
I have defined Magnetic and Solid Mechanics physics, but I want comsol to calculate the magnetic force and put it as boundary load in solid mechanics.
Because if I put the magnetic estimated load (magnetic module estimate) as body load in solid mechanics, then the force direction is decided by me, and not by Comsol.
So this is not actually coupling, I am getting the force by magnetic module estimation and putting it as boundary loan in solid mechanics and deciding in which direction it should be!!
I would appreciate your help in this matter.
Thank you
I have attached a model. In this model I have two magnets which can be the same polarity or opposite.
The magnets are in tip of the rods, and I want to see the rods deform when the magnets have same or opposite polarities.
I have defined Magnetic and Solid Mechanics physics, but I want comsol to calculate the magnetic force and put it as boundary load in solid mechanics.
Because if I put the magnetic estimated load (magnetic module estimate) as body load in solid mechanics, then the force direction is decided by me, and not by Comsol.
So this is not actually coupling, I am getting the force by magnetic module estimation and putting it as boundary loan in solid mechanics and deciding in which direction it should be!!
I would appreciate your help in this matter.
Thank you
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hello Mr. Yushanov,
I have been trying to do the same with 2D model witth square shapes. I implemented the familiar steps but result is failure. Can you help me to solve this?
Sincerely.
I have been trying to do the same with 2D model witth square shapes. I implemented the familiar steps but result is failure. Can you help me to solve this?
Sincerely.
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
At most I can get NaN value :) and no motion in animation.