Discussion Forum

Hi

Are you sure the relative magnetic suceptibility of steel is "1" ?, I would rather use a sot iron with mur=4000 or something like that, I believe that would give you some more realistic results, no ?

Like that for the B field, air and steel are rather equivalent, no ?

Often its better to not use the default material values, and fill in by hand only the one required, then you would have a better control of what you are doing, and also its easier to debug

Have fun comsoling
Ivar

Thanks you for your response, Ivar. I followed your advice, and understood why I couldn't see the current. In effect, when I enter my own parameters for the material, I can see the current density in my bar !

Here is, I think, my last question in order to manage to finish my simulating work: the ideal would be to visualise the changing impedance of the "coil" when the cracked bar is translated in it. Is it possible to do so with COMSOL ? Can i do that in the Results/One plot 1D or must I introduce an Eletrical Circuit than allow me to draw Reactance=func(Resistance) [impedance complex plane] ?

Thanks again

Hi

I'm not sure how to do it in V4 yet, will try, the impedance is basically the voltage over current, but you should expect it to bne compelx with real and imaginary parts (phase) as you would scan the frequency with a parametric sweep. Your current deault frequ=100Hz is probably a little low in value (all this if I find some time).

Now, one thing puzzles me: with the gradients you have and the very coarse mesh you use, you are currently highly mesh dependent, I would use boundary mesh elements with high density along the borders, as you have surface curents all around, you will notice how the field lines changes with the mesh. you could also try wit PML (if implemented yet to reduce the total air volume meshed)

In V3.5 you could use assembly mode and move the iron part along Z, currently I do not know how to do it in V4 (it's the same principle as for the rotary generators from the COMSOL examples, there is also a few models on the forum from some previous linear motor examples, try a search

Have fun Comsoling
Ivar

Hi

A few other things:

You use the variable name X for your Z position, it might well come in conflict with COMSOLS internal names such as x,y,z X,Y,Z ... I would prefer to not use names known as internal COMSOL names, so why not "Zpos" ?

There are some nice new coil BC in V4, you can use a "singe" (or multiple) turn coils and have an excitation in V which allows easier the calculations of the current and then to take the ratio for the impedance.

But I have also discovered some incoherences in the probe plots of the current V4.0.929 with the units between average and integration, just as well how these generate solutions. And I have some problems to regenerate a sequence if I delete the current solve sequence, cannot understand why. Some of these I will report to "support" as I would appreciate to understand especially when it hits my understanding of formal math/physics ;)

more as it arrives

Have fun Comsoling
Ivbar

I changed X into Zpos.
In fact I want to draw V as a function of Zpos, and the current in the bar in fuction of Zpos too. Then I can have the ratio and the impedance.

Are you saying you know how to do it in 3.5 ? Is it easier ? Because if you're sure I can manage to do that in 3.5, I can ask a collegue of mine because he has got 3.5 version.

By the way, what do you mean by "there are new coil BC in V4" ?

Hi

If you look carefully, there is, in V4 the single turn coil and multiple turn coil boundary conditions (BC) these simulate coils in two different manners, if you look carefully a single coil at higher frequenceies have significant geometric effects due to the skin behaviour of HF currents (even at 1kHz) that are pushed to the coil surface. And you can define current or voltage excitation.

With multiple turn coils you get a more homogenious current density, and it has incorporated induced voltage interactions. Check that you have the update V4.0.0.929 and look at the acdc.pdf doc

in V4 it should be as in V3.5 for the rotary/linear displacements. You can either do a static motion of the slit, as you are doing (but you need to add a parametric sweep in the job sequence, add it in the study and have COMSOL to generate the sequence, then set Zpos as parameter). This will do a static=fixed geometry modal analysis of your case (which is probably slightly different if your slit was moving with a constant velocity). The other way is to split

To implement the moving Mesh (with constant velocity effects) I know I have send a model on one of the threads a few months ago but I cannot locate it, I have answered too many cases ;) But basically you must split your air volume vertically about mid-way between you iron and the coil, then compose the geometries to have two composed items, one fixed one moving, and use assembly mode (geometry Finish option) then you can use the moving mesh with a velocity. It is basically repeating what is in the 3.5 doc ACDC about the rotating generators, but arranged for linear motion.

Finally I would use PML to limit the external "far away" boundaries to a smaller volume, and increase significantly the mesh around the iron (use iron and not steel as material properties) including boundary mesh elements, as you have significant skin filed gradients.

Have fun Comsoling
Ivar

Ok, I will study and try what you told me.

I keep you in touch with my results by tonight, editing this post, or replying.

Thanks again, Ivar.

Ernest D.

I haven't tried your solution yet. Before that I have few questions maybe you can help with:

- When i replace the material Steel AISI by Iron, an error appears when I try to compute the solution (Failed to evaluate expression). Why ? :-) The same problem appears when I choose other materials than Iron.

- Is there a way not to put air in my problem ? Because I want to visualise the induced current in 2D, but in my surface plot, the color of the air correspond to high currents, wich is a bit illogical. I think It triggers incoherences with current in my bar because they became higher with detph.... They should be located at the surface :-(

By the way, sorry if my language is fragile, I don't speak English very fluently.

Thanks again

Hi

have you checked the material properties ? are there any "red STOP flags", as iron is normally defined with a mur=4000 but if I remember steel and other oftehn do not have all magnetic properties defined.

personally I prefer to define the material properties by hand, then I know exactly wht is in there, and mostly there are too many ddata and they are often non-linear hence longer to solve without necessarily any better information.

But once you have your model and you are confident with it, its often funto set more detailed material data in, and make a final run to see the differences (mostly negligible) but that depends alot on the type of model you are doing.

have fun Comsoling
Ivar

Hi

I got a very usefull hint from support.
When we add two or more physics and solve with a common study that is not the ideal one for both/all, but still want to link the study to solve both all physics in one go, we must use the Physics Magnetic field (in your case) Advanced Settings - Equation form: Frequency Domain - User defined - 50[Hz], then select a time study, apply to both physics (but magnetics will be forced to FD) and off you go.

Still why does FD reset T to 0[K] ? I will keep this one for next meeting with the local Comsol rep, its in a few days ;)

Again: the higher up in the tree you define something the more is fixed for further down, but this internal link is easy to forget, if you open the file again in a months time

Have fun Comsoling
Ivar

Hi Ivar,

I am simulating a Thomson repulsion drive, somehow it is a similar proble to the one discussed in here. My problem is the following:
I want to discharge a capacitor bank through the coil in order to generate the repulsion on a moving plate above. But in PSPICE I didn't manage to do it since there are no breakers, capacitors don't allow precharging and resistors cannot be variable. I also tried to transform the model into a Simulink S-function but the simulation becomes unstable.
Does anybody have a suggestion?


Thanks!
Josep