Discussion Forum

Hi

can you not try out by going to the physics settings of your surrounding subdoman and in the "constitutive relation" select either B=mu0*mur*H+Br or B=mu0*H+mu0*M and define a Br or a M in the desired direction and of the desired strength ?

By the way, be carefull with interpreting the Maxwell stress tensor values, check that they seem coherent as they are rather mesh dependent because they rely on a good precise and coherent boundary integration

Good luck
Ivar

Define the space where your particles will move (e.g. a cube in 3D) and specify the field intensity at the boundaries, that way you don't need to worry about the field sources .

Setting the field at all boundaries works for creating a uniform field region once you realize that the "Magnetic Field" boundary condition specifies the "tangential" component of the field in absolute coordinates -- a somewhat non-intuitive convention (at least to my mind).

So, for example, if you want 1 A/m field in the z direction, 4 sides of your boundary cube will specify 1 A/m z component, wheras two of you cube sides will specify zero z component!!!!!

Hi,
I am using comsol 4.2a. I am trying to simulate the hartmann flow problem. I have two parallel plates and a conducting liquid is flowing through it, in the presence of uniform constant magnetic field. I do not know how to produce this type of magnetic field. I am working with Laminar flow(spf) and Magnetic fields(mf) in 2d geometry.
Can you please tell me how to produce the uniform magnetic field.
Thanks
Mithlesh

Hi

There are several options, but first of all so you really need MEF and solving for V ? (haven't checked the application though) as its far heavier than MF and often MF is enough at least to debug and start with. (in MEF all materials must have a conductivity different from zero, except if defined as a Ampere laws domain)

One way is to make the surrounding a cylinder (add a cylindrical coordinates along with the cylinder axis to define the current more easily) and set a phi current flow in the cylinder shell, another way is to set a top and bottom. But because of the vector field behaviour of the Maxwell equations you will get non uniformities along all sharp edges, so often its worth to cut out the outer tube in the vicinity of your current shell (add a 10% radius layer and hide the external tube part in the view hide entities)

But this solenoid type fields is probably not "uniform enough (but its rather realistic of what one achieve from an experimental "uniform" B field)

So to get it "more linear", try adding a "Perfect magnetic conductor" on the flat top and bottom surfaces of the cylinder
remains to ask yourself how "physical" this is, I leave that to you

the attached file is 3D and v4.3a.161
--
Good luck
Ivar

Hi Ivar,

The method to apply magnetic field is very effective and common in the comsol module cases. I tried to use this method before but the results cannot give me the UNIFORM magnetic field. If I add the remanant magnetic field in the domain, it simply change the domain as a permanent magnet. And I think this is how comsol calls that as "remanent" term.
May be my claim is not exactly what you wanted to say but any comments or suggestions would be greatly appreciated.
Thank you very much Ivar.

Yi

Hi Juan,

I use the method exactly the same with your description. But the magnetic insulation will be overridden, to correct that, I build a bigger sphere outside and force the magnetic insulation condition there.
The results shows me the uniform magnetic field through the whole domain where I add the magnetic field which is good. But the other results are not what i expected.
On the contrary, if I apply the magnetic field as the remanent term in the whole domain of the moving particle, instead of applying on the boundary, but keep anything else the same. then the magnetic insulation condition is well defined and I can get the close results (like velocity contour, pressure drop...)with the analytic solution but then the Magnetic field would't be UNIFORM.
Do you think where I may make mistakes?
Thanks

Yi

Hi Chris,
Good to know that, thank you very much for this trick.:)
But I have one more question. If I apply my uniform magnetic on the boundary as the "Magnetic field boundary condition" then I will lose the magnetic insulation boundary condition right? And There is no problem of losing the magnetic insulation in my model when my Hartmann number is low, but If it's pretty high, like 10,000, then I will give me an error called "divergent cleaning failed" something like that. That is because the comsol cannot get the magnetic insulation boundary condition.
Do you have any idea about that?
Thank you very much

Yi

Hi Mithlesh,

Did you get what you expected? I am also doing the simulation of Hartmannn flow by comsol but I am not sure whether the way I apply the uniform magnetic field is correct or wrong. I described my ways here in series questions.
Thanks,
Yi