Discussion Forum

Hi

but can'nt you couple the physics so you get the deflection of the membrane and you estimate the capacitance from the EC calculation on the defomed shape ? That is a question on how to segregate/link the solvers

--
Good luck
Ivar

Hello Sir Ivar,

We can sure integrate both the physics. Only problem is i'm not able to link the two physics. How do i do it. Could you help me out. Any modules/documentations i can get on this?

Thanks.

Cheers!

Abhi

Hi

the basics is to couple the structural deformation to the EC or ES calculation. This is mostly done via ALE. YOu should have the example of the capacitor accelerometer (also including squeezed film, but start with the simple example) For 3.5 I believe you have also a few models in the Model Exchange page of the Community

--
Good luck
Ivar

Hi

Thanks a lot on the varying capacitance guidance Ivar. Helped a lot, worked perfectly. But only concern is this coupling two physics together... I could not find this capacitive accelerometer u pointed out.
However i tried the squeezed film method...Unfortunately it is not working for pressure sensor.

Any way i can look up the coupling of different physics environments so that i can implement it on the sensor?

Also, a MAJOR problem Sir Ivar.

The deformation.I use version 4.0a. In the example that i've hereby attached, the gap b/w the plated is 10um and deflection is 3.16um. But in the graphic window it shows that the deformed top plate is surpassing the domain of the bottom plate!!! I mean thats impossible rite? any reason as to why it is happening?major issue it is...

thanks

Cheers!

Abhi

Hi

normally COMSOL detects the different physics and proposes links when it understand potential complementary entries (not working perfectly yet in v4.0, far better in v4.1, hopefully perfect in v4.2 ;)

Whenever else you must link the physics "by hand" using the equations from one as input for the others, do not forget to couple both ways ;)

Then for crossing inter-penetrating geometries do not forget that by default FEM has no indications about contact and collision, so you can get "legal" valid results that are not physical because of collisions in the real world. What I suspect though, from the dimensions you state, is that you have not noticed that the "deformation" scale is set by default to "auto" which is a value to give a nice picture. Check the Deformation node, and select scale "1" and you might get a more realistic image. Do not forget that normal deformations to stay within elastic limits have < 0.2% strain, this implies often so small deformations that it does not show up on a screen, therefore the "automatic, default scale adjustments
--
Good luck
Ivar

Hi

YES! Got everything. But while trying different stuff i see that is i change the 'Geometric Entity Level' of the 'Integration' in coupling, the unit for the same formula changes in the global definitions.

like for varying voltage based on changing capacitance i used the formula

v = es.Q0_1/C {here,C = mod1.intop1(1*elsilon0/w)}

i defined new integration and variable intop2 for defining voltage.

Now, for C, i chose Geometric Entity Level as boundary in intop1.

if i choose the same, i.e boundary, for voltage for the above formula, the unit for V is V.m^2 !!! and the result too is erratic.

How can i resolve this issue?

Cheers!

with lots of thanks again

Abhi

I found this model of capacitive pressure sensor. I looked into the 3D model. And surprisingly no es module has been adapted. How can u get the output voltage without that then!?

I mean, to detect the change in pressure we need some means and it can be done only by finding the voltage change w.r.t the changing capacitance rite?

Some help on how to formulate this please...

cheers!

Abhi

Hello,

Also, for the es physics to calculate capacitance, it needs a dielectric between the plates! how to provide vaccum as dielectric then!??
Because i thought of deriving the charge from es physics and integrating to get voltage from the formula
es.Q0/C where C is the capacitance w.r.t deformation for corresponding pressure.

Now, since i dont keep a dielectric, charge and capacitance are of the order 10^(-39) instead of 10^(-12).

well......??

Thanks

Abhi

Hi

you have many questions, I see, and some are related with how COMSOL works. normally most of the variables are "densities" per m^2 (for fluxes) or per m^3 and apply to the mesh elements of size dx*dy*dz (in 3D) so you need to integrate these values over a Domain (dx*dy*dz) or a Boundary =(3-1)=2D) dx*dy, or dx*dz or dy*dz or whatever coordinate system you use.
This multiplies the densities units by a m^3 respectively m^2. That is why the integration changes. This is pure math/physics no black magic ;)

If you are in 2D, you should not forget the default "depth", normally 1[m] but in certain physics you might overright this, so you need to add a i.e. "solid.d" in your integration to get the right units, or you read them out as "per meter depth"
In 2D-axi, you have the loop length, that you might add in automatically in certain integrations by the advanced tick include 2*pi*r

Understand these notation issues are essential to use correctly COMSOL. My advise: always check the units, and incase of differences with what you expected think it over find out why, it will soon become a habit
--
Good luck
Ivar

Hi
I have the same problemlike abhi. I am trying to find capacitance from vibrating parallel plate capacitor. My structure is same like abhi. But for my capacitance calculation I wanted the the gap(x, y) will be changing as because both plate is vibrating. I know how to make integration for one plate but i dont know how to make integration for this type of variable epsilon/(D1-D2+0.5). Where D1 is displacement w for upper plate and D2 displacement for lower plate and 0.5 is the static distance between both plate.How I can do that?

Thanks

Hi

you need to map the values (u,v,w) from one boundary/edge to the next one so you can then use bth local and remote displacements. Look at the Definition coupling variables

--
Good luck
Ivar

Hi Ivar
Thanks for the suggestion. I am working on that. I have another question. Previously I tried to find the capacitance directly (using AC/DC-electrostatics module). I tried to make a block of air and put my sensor inside the block. (This method I learned from one library model, they tried to measure the static capacitance). But in my case I am doing a time dependent vibration simulation where both the upper plate and lower plate is vibrating, and I want to measure the capacitance at each time step. The simulation does not work because they want me to put the value of Poisson ratio and Young's modulus of air. I don’t know how I can define them for air or vacuum. Thus I have three questions
1. Is it necessary to make a block of air for the dielectric simulation?
2. If it is necessary is it possible to do structural mechanics simulation (vibration) without the block of air and add it for electrostatics simulation.
3. How I can define Poisson ratio young’s modulus for air?
Thanks

Hi Mustafa,

Firstly, I would like to precise that you cannot solve your problem by integrating the work of the electric field for the area of electrodes. It is only possible if you have constant gap between electrodes.
In your case I think you defined on both plates (or just one, it doesn't matter) forces, which oscillate giving a similarly varying gap between plates. Since the gap is varying you cannot apply simple mesh elements. You should add physics to your model, then choose "Mathematics", "Deformed mesh" and "Moving mesh (ale)". (ale=Arbitrary Lagrangian/Eulerian mesh). The ale mesh will allow the model redefining the mesh geometry according the displacement in the air-gap. In the moving mesh you should apply "Prediscribed mesh displacement" and set dx=u and dy=v if you have a 2D problem (and dz=w if 3D) for the boundaries of the electrodes contacting the sendwiched air-layer. Then you can mesh your structure. You can find the model www.comsol.no/community/exchange/118/ to have some more ideas.

Concerning the other questions:

I don’t know how I can define them for air or vacuum. Thus I have three questions
1. Is it necessary to make a block of air for the dielectric simulation?
You absolutely need the air or another dielectric defined between the electrodes if not there is no dielectric displacement, no capacitance.
2. If it is necessary is it possible to do structural mechanics simulation (vibration) without the block of air and add it for electrostatics simulation.
To solve the problem is simple. You should apply coupled physics in your model, add both "Solid mechanics" and "Electrostatics". You should also add "ale mesh" , thus there will be 3 physics in total. You should unselect the air or dielectric in the "Solid mechanics" tag, just select your 2 electrodes. Then in the "Electrostatics" choose the air layer only.

3. How I can define Poisson ratio young’s modulus for air?
It is not relevant to your problem anymore as these properties are not considered anymore in your model if you followed the above settings.

I think I didn't forget to mention anything important.

Good luck.

Robert

Hi

in V4 with the spatial frame you can also "cheat" and not use ALE (if the deformations are small), you give a density and young modulus + Poisson to the air, sufficiently small to avoid influencing the structural part.

Then you get the deformations correly. This kind of models are easier to set up than a full ALE.

But the "correct complete" way is, I agree via ALE

Note forget 4.0, use the latest version, it's far more stable

--
Good luck
Ivar

Hi
Thanks for the very informative message. I tried to add three physics (Solid mechanics, Electrostatics and Moving mesh(ale)) in my simulation. But it didn’t work. Probably I have mistaken on some boundary condition. My questions are
1. Will I have to add three physics at the same time and then mesh?
2. Will I have to add three studies or one study will work?
3. Previously I was doing only the solid mechanics simulation with a “prescribed displacement” then it was working but now an error message came “failed to find consistent initial value”.
please take a look at my file