Discussion Forum

I am new to comsol too and ran into this problem yesterday. I found in an old text of mine that the permittivity of a conductor is that of free space. This is because permittivity is caused by the polarization of fixed molecules. So I entered 1 for the relative permittivity of the conductor.

The way to do this is to just not solve electrostatics inside the volume of the metal, and apply boundary conditions (such as fixed voltage) on the metal surface.

Hello

I'm completely new to comsol. I'm trying to simulate the electric field between some specifically arranged metal electrodes. Between them is some material with a certain permittivity.

Now I get an error because there is no permittivity (epsilon) defined for the metal regions. Of course the metal part should have theoretically infinite permittivity. So I tried entering a very large number there but I'm not sure that this is the proper way to handle it. Is there some other model that should be used for simulation of the metal parts?

Hi,
I want to ask how you defined your boundaries. I am also new to COMSOL and need help with boundary conditions for a similar problem.

Hello

Thank you all for your contributions. I am assuming that not solving electrostatics in metal regions is most probably the way to go. afaik, if you take '1' as the value for epsilon you obviously ignore the field-screening effect of the conductor, so this cannot be the solution. My problem now is that I have some floating metal parts (floating as in not connected to the ground or any other voltage source for that matter). Therefore I cannot define a voltage boundary condition because this voltage is exactly what I want to find out.

Hi

check carefully the example with the "floating electrode" not sure it is documented in V4 yet (could be) but it was in 3.5a. This gives an example of how to define a floating potential electrode and the get the resulting value out (basically by adding an unknown variable to solve in paralle with the rest).

Good luck
Ivar

Yes you are right, you don't need to include the metal as a domain, take it just as a boundary, if your assumptions do not include its real physical property. (because even taking it as a boundary is an approximation!!).

If the electrode is not connected to any power supply, this means that it should be a its OCP :open circuit potential. The potential that should be on your electrode depends on your assumption that you are considering in this study, mainly the charge density in the adjacent domain.

As Ivar said, in this case, you have the floating potential boundary option, which works quite good!!!
This means that at this boundary none of potential and surface charge density are fixed!! Your electrode in this case is polarizable!! The final potential at this electrode can be affected by the adjacent domain (charge density, epsilon..) and of course by the floating potential that you assumed.

Considering the epsilon equal to one for your electrode is wrong !! at least the same epsilon as the adjacent domain would be more correct.

I hope it helps

Cheers

Yes you are right, you don't need to include the metal as a domain, take it just as a boundary, if your assumptions do not include its real physical property. (because even taking it as a boundary is an approximation!!).

If the electrode is not connected to any power supply, this means that it should be a its OCP :open circuit potential. The potential that should be on your electrode depends on your assumption that you are considering in this study, mainly the charge density in the adjacent domain.

As Ivar said, in this case, you have the floating potential boundary option, which works quite good!!!
This means that at this boundary none of potential and surface charge density are fixed!! Your electrode in this case is polarizable!! The final potential at this electrode can be affected by the adjacent domain (charge density, epsilon..) and of course by the floating potential that you assumed.

Considering the epsilon equal to one for your electrode is wrong !! at least the same epsilon as the adjacent domain would be more correct.

I hope it helps

Cheers

Yes you are right, you don't need to include the metal as a domain, take it just as a boundary, if your assumptions do not include its real physical property. (because even taking it as a boundary is an approximation!!).

If the electrode is not connected to any power supply, this means that it should be a its OCP :open circuit potential. The potential that should be on your electrode depends on your assumption that you are considering in this study, mainly the charge density in the adjacent domain.

As Ivar said, in this case, you have the floating potential boundary option, which works quite good!!!
This means that at this boundary none of potential and surface charge density are fixed!! Your electrode in this case is polarizable!! The final potential at this electrode can be affected by the adjacent domain (charge density, epsilon..) and of course by the floating potential that you assumed.

Considering the epsilon equal to one for your electrode is wrong !! at least the same epsilon as the adjacent domain would be more correct.

I hope it helps

Cheers

Yes you are right, you don't need to include the metal as a domain, take it just as a boundary, if your assumptions do not include its real physical property. (because even taking it as a boundary is an approximation!!).

If the electrode is not connected to any power supply, this means that it should be a its OCP :open circuit potential. The potential that should be on your electrode depends on your assumption that you are considering in this study, mainly the charge density in the adjacent domain.

As Ivar said, in this case, you have the floating potential boundary option, which works quite good!!!
This means that at this boundary none of potential and surface charge density are fixed!! Your electrode in this case is polarizable!! The final potential at this electrode can be affected by the adjacent domain (charge density, epsilon..) and of course by the floating potential that you assumed.

Considering the epsilon equal to one for your electrode is wrong !! at least the same epsilon as the adjacent domain would be more correct.

I hope it helps

Cheers

Thank you, this example was very helpful!

According to the example there are two possibilities: either define a terminal with zero current (I = 0) as boundary condition, or a electric shielding boundary condition. These are only available in the electric currents ("ec") module. But when I change from electrostatics (es) simulation to ec simulation I have the same problem as before: All the parts made of dielectric material cannot be solved because the sigma (conductivity) value is not defined there,

In general I get an error that epsilon is not defined if I use es and sigma not defined if I use ec. Does it make sense to simulate the insulating parts with es and the conducting parts with ec?

Hello Clemens,

I was wondering if you figured out the problem you were having with not being able to apply a sigma value to dielectrics and vice versa. I am doing a simulation very similar to yours and have been having a lot of trouble applying boundaries/subdomains and meshing.
Thanks,

Michael

Hi

have you tried to apply a conductivity of 1[S/m] ?
often one can get around with just a low "insignificant" (but not innocent) conductivity
--
Good luck
Ivar

Hi,
@ Ivar: You have mentioned about a 'floating electrode' model. I haven't found any such model (neither in 3.5 nor 4.0). Please, give me the exact name of the model or better send me some online link etc.
@ All: I have to calculate the force of attraction between two metallic blocks. Both the blocks are on top of insulators and one of them can be charged using a charged rod. (I am using Comsol 4.0a>Electrostatics>ACDC>Stationary)
1. How can I find if an equal and opposite charge is actually induced on the second block? What boundary conditions should I apply on the two metal blocks? (Currently, I am myself specifying an equal and opposite charge on the untouched metal block using floating potential/ terminal boundary conditions on the two metal blocks).
2. The epsilon r (dielectric constant Er) for metals is given as 1 in the material browser. I just tried Er=1+9999j and there was no error but the force calculation had an imaginary part as well (I am not sure what the imaginary part of force means). Is it right to use such an imaginary no. for Er? If yes, what should be the value of the imaginary part? (I used 9999 only to specify a large imaginary part).
Please, see the attached file for details.
Waiting for your reply.
Yours truly,
Abdul Rehman Javed.

Dear Clemens,
Please, mention the example that you are mentioning about. Send me the link or the exact path where I can found this file. Your problem is quite relevant to my problem and I shall like to know what you finally used in your case. Please, reply.
Thanking you,
Abdul Rehman Javed.

Hi

I have taken a look at your model and I have a few comments:

1) upgrade to the latest version (see your rep or sysadmin) as the early 4.0 as many weaknesses that make it rather frustrating to use
2) you have not defined any epsilonr for the "air" material (always recheck your materials when you have finished your model set up, as depending on the options in the physics you might need new material properties, you have currently a "red flag" shown there!)
3) your air gap between the two items is very small and its "connected" to the rest of the air, you should ensure that you have 3-5 elements across this gap, as this is where the details of the gradient will apear. There are different means to achieve this. Personnally I "cut" my "small" regions with a rectangle too, adding a small interiour boundary but that does not really change anything for the calcualtion speed. Then I can apply a very fine mesh in the critical recion, and leave COMSOL do the rest. In anycase you should edit the default settings of your mesh using the feature "Size - Custom - Resolution of narrow regions" = 3 or 5
4) then you have a very complex epsilonr for aluminium, for me its rather =1 and not 1+9999j but, I do not know what you are testing, exactly.
5) the resulting forces are very high but again you have a very narrow region and quite some charge, compare it to hand calculations, as with the high gradient you might get rather large errors on the forces due to this gradient and the mesh resolution.
6) as you system is rather "round" I usually use a circle and not a squre of air around my systems, and I use iinfinite elements to reduce the extra air size and increase the mesh in the critical region
7) as you case is fully plane symmetric, you could simulate only the right half, hence increase further the mesh density
8) your depth here is 1[m] the default value. But as your parts are a few microns, you have a scaling factor for the Z direction size to consider too (at least not to forget ;)

--
Good luck
Ivar

Hi Ivar,
Many many thanks for your reply.
1) Is it a free upgrade? I shall ask the admin in my university but I am not sure if they will do it for me.
2) I created this simpler design for posting here so I might have missed the epsilonr for air. Thanks for mentioning.
3) By ''cutting'' into ''small regions'' do you mean to say that I should make a rectangle (of air) in the gap between the two Al plates? I was using this technique earlier and shall reuse it, after your recommendation.
4) *** This was actually my main question. In the discussion above and in other texts, metals have a complex (rather purely imaginary) dielectric constant dependent on frequency of incident/ exciting em waves. In the discussion above and in some models they actually do like deleting the metal block from the enclosing air box and just define constant voltage boundary condition on all faces of metal plate. That might work for a capacitor simulation but this is not what I am intending to solve. Should I use the default value of the epsilonr i.e. 1 in my case? Please, elaborate this. Note that this is just equal to the epsilonr for air.
5) The forces are really unprecedentedly large because of the very high charge that I have placed on the plates. I am not sure how large a charge can I place on the Al plates actually so I was just using this fictitious value. Do you have a clue as to how large a charge can I place on the Al plate (by touching with a charged rod)? I just need a rough estimate.
6) I haven't used the sphere/ infinite elements technique so far but will surely look for it.
7) I am not sure about plane symmetry. I have used axial symmetry before but I haven't heard about plane symmetry.
8) In my actual field the depth field is also in micro meters.

Please, do comment about points 4 and 5.
One last thing, since there is no way to actually induce charge on the right Al plate so I am using the following boundary condition on the right Al plate: (-Q) charge on the left wall and (+Q) charge on the right wall (assume I have +Q charge on the left Al plate). Is this right in your opinion?

Hi

I would say one shall always consider keeping a maintenance on such important simulation software, then upgrades are "free" ;)

well your case is a static analysis, so I would say epsilonr=1 should do.
If you want to do an RF analysis at high frequency by a freqeuncy scan, you need to consider skin effects etc and appropriate material data, either by a complex epsilon, or n-i*k, or other means as you state

Yes I would split the air gap with a rectangle (three rectangles side by side, one small one for the air gap.

Plane symmetry: you cut your geoemtry in the middle of the air gap and apply a SYMMETRY BC on this line. It's similar to 2D-axi, but in done standard 2D: this means that you are NOT looping around the axis, as for 2D-axi, it's just a virtual plane going out of the paper (along Z both directions) in 2D

Maximum charge on a round plate: dry air allows some 1kV/mm electric fields at athmospheric conditions and no sharp corners !! So you should be able to estimate the total charge, taking into account your geometry = capacity (not forgetting the depth dimension = 1[m] if you are not changing it along "Z" .

In vacuum it changes, but you are going through the "Paschen curve inversion" (around 10 mbar) for breakdown when vacuuming out, as the discharge voltage is linked to the mean free path of the residual electrons, and atoms/ions

Note: COMSOL always consider 3D, in 2D it sets mostly Z=1[m] and simplifies the equations accordingly

--
Good luck
Ivar

This might be old, but is still a very useful thread to me as well. On paper, its easy to calculate the electric field outside a charged spherical surface, without assigning an electric permittivity to the surface "material". In COMSOL, I was really wondering deeply how to do the same thing for learning purposes, but couldn't manage because it was unclear to me, what the permittivity of said surface should be. So the electrostatics are not solved for the surface, rather it is considered a boundary.

I am drawing the basic diagram and understanding the software in detail. My research work is in the cables terminations, joints and electric field.

I have draw a simple two electrode rod in the software with First rod at 0V and second rod on 1V. The material property i select the copper and air for the best results. But when i run the electrostatic analysis on this modal, there is a equi-potentials lines in the copper with permittivity of 0.99996 and in air with permittivity of 1 (air region is fine). According to theory, there will be no equi-potentials lines in the copper.

If you have a boundary condition around a region and the permittivity is low (that means insulator) you will have potential lines inside the region.

But, if i change the permittivity of the material (mean high), then there is no equipotential lines inside the copper region and there is no problem in the design. but it mean is that, it is not the same material i am using. Example: my design is on power cable (copper material with any insulation material) and relative permittivity of copper is =0.999999666. and free space permittivity is =1. so E=E0xEr is the equation. That mean E=1. if i change the permittivity, it mean i am not using copper. I am also using the potential with the permittivity.

It is not supporting the theory and mathematically equation. Kindly, guide me with this problem. Thanks

i am new user of comsol. i want to model explosion of dust. the source of the ignition is 2 electrode with 6mm gap. but after a certain time this electrode can ignite. does anybody know which physic should i use??
thanks in advance

@Ivar Kjelberg I am working on COMSOL to simulate vertical contact mode Triboelectric Nano generator in which by changing (increasing) the distance between two surfaces, the potential difference increases. I am using PDMS and copper as two surfaces and the charge density on both surfaces is 8 uC/m2 and -8 uC/m2 respectively. I want to use ES module, 2D and Stationary parameters.
Thanks in advance