Discussion Forum

Hi

I would suggest to try the following: start in 2D, once you understand the principoles fully you can go to 3D, then if you hve a top and bottom plate with some air in the middle and a pressure load on the top and bottom plate, while these are held fixed by the lateral edges, you will see the plate compress, and the average distance between the plates change. If you assume these form a capacitor, with the spacing change you will expect a capacitor change.

Start 2D physics annd in 1) ALE (FIRST important) 2) Structural 3) ES physics, select stationary solver case and Finish. Then go to the geometry, draw a rectangle, lets say 10 mm wide/long 2 mm height, go lower down and find the Rectangle "layer tab" and open it, add 0.2 .. layer on top and on bottom. Jump to the material node, add two materials, lets say Built in Aluminium and leave the default for all domains, then add "air" and select the domain 2 in the middle.
Leave the ALE physics for the moment and open the Solid physics to be shure it applies only to the top and bottom domains and add a fixed BC condition and select the short edges left and right of the top and bottom plates. Then add a Boundary load, lets say 1000 N/m^2 pressure load, and select the top and bottom external boundaries of the respective top and bottom plate. This will provide some compressive load to bend the plates.
Now go to the ES, select the middle domain only, set a GND BC on the lower air side (plate - Air interface) and add a terminal or fixed voltage BC on the upper air - plate interface.

now jump back to the material and check the two nodes, are all material properties defined, you might need to add something, like en epsilonr for the air, use then "1".
Then only move to the ALE physics. You will notice that only the air is concerned, as the top and bottom palte is overriden by the solid physics (solid contain something ylose to the ALE with their differentiation of the spatial and material frame, this is why ALE should be FIRTS in the physics list). Select a doamin free deformation and select the air domain. By default all boundaries have prescribed displacements "0" so you need to add now a new Prescribed displacement node, sleect the two top and bottom plete to air interfaces and use u,v (resp w in 3D) as prescribed displacement values for these boundaries. U,v and w are part of the solid dependent variables, and as the boundaries are common to the air and solid, they heritate common variables. This law will enforce that the ALE boundaries follows the solid deforming paltes.
Now its "just" to solve and observe the results.
For the capacitor you can get them if you have the right modules (MEMS or ACDC) via a Derived Value Globa Definition terminal capacitance, else you must calculate the capacitance by performing an integration on the plate bound0ary of the separation of the two electrodes applying a standard capacitance formula

Hope I didnt miss anything ;)
--
Good luck
Ivar

Thanks Ivar for the reply.

I tried your step by step instruction. But I got the error message saying that

"Evaluation of plot mesh vertex coordinates failed."
"Failed to evaluate variable.
-variable: mod1.u
-Geometry:1
-Domain:2
Failed to evaluate variable.
-variabel:mod1.ale.dxx
-defined as: mod1.u
Failed to evaluate expression.
-Expression: X+(mod1.ale.dxx)
-Plot:surf(Surface)"

In the ale physics, i added "free deformation 1" and then added "prescribed deformation 1". One thing I noticed that was different from your instruction was that for the ale physics setup: instead of "add now a new Prescribed displacement node, sleect the two top and bottom plete to air interfaces and use u,v (resp w in 3D) as prescribed displacement values for these boundaries.", I was only able to select DOMAIN 2, not the two plates.
Why would you have the "free deformation 1"with domain 2 first and override it with "prescribed deformation 1"?
What is "free deformation 1" for?

In the Study1->step1:
all three physics are checked and in the order of
ale
solid
es

It looks like the ale can not see the mod1.u during the study. Are they sequential?

Thanks again.

Best,
Stacey

Hi

By default, for ALE, COMSOL proposes a fixed domain mesh and fixed = prescribed to "0" for all surrounding boundaries.

Be sure the ALE applies only to the domain (=volume in 3D or surface in 2D) that you want to deform the mesh for, and NOT to the solid domain (perhaps v4.2 is somewhat different ?)

Then be sure the ALE is before the solid in the node tree list

Now we can define the ALE physics:
Add a free DOMAIN node and select the domain to deform if I read you correct you have only 1 domain
Add a Prescried Displacement BOUNDARY node, select the common boundary with the solid and set x,y (in 2D, and also z in 3D) to u,v,w respectively

This works for me ;)

--
Good luck
Ivar

Hi,

When I right-click Moving Mesh, there are these options:
Fixed Mesh
Prescribed deformation
Free Deformation
Prescribed Mesh Displacement
.etc

There is neither "free DOMAIN" nor "Prescried Displacement BOUNDARY".

Thanks.

Best,
Stacey

Hi


you should have ome icons, the first series are for domains, then a line will separate the boundaries, lower the edgs and last the points

Take a look its 4.2a

--
Good luck
Ivar

Hi,

Now I see what are for boundaries and what are for domains. There are the default "Fixed Mesh"(for domain) and "Prescribed mesh displ"(for boundaries) generated with "ale". I can neither modify those nor disable them.

When I add another "Prescribed mesh displ" after "Free Deformation", I can't choose any boundary since they are all (not applicable) just as in the one defiantly generated...

Do you see where my problem is?
Thanks.


Best,
Stacey

Hi

when you load in a physics, there are a few default nodes that comes with for domains and for boundaries, such that the physics, by default, will be coherent. Tese have a little D in the icon in 4.2a and are always the first nodes, and cannot be deleted, BUT they can be overwritten.

In the main physics node you select the domains that apply to your physics, in the lower subnodes you select all or only some of the domains and corresponding boundries per subnode, BUT you must ensure ALL boundaries of the main physics node are defined at least ONCE, and that enough BC conditions are set for all subphysics to give an unique solution of the PDEs.

Therefore you will see some overrides appear depending on how you combine you node and in which order (top/down) they appear.

If you can add a free DOMAIN, and a Prescribed Displacement Boundary sub node, but cannot select any domain or boundaries, then it means that something is blocking you, from the ALE physics or from another physics.

One important fact: the ALE physics should appear BEFORE (=top) the solid physics node (=bottom) in the node list (note: this is not well documented today, COMSOL has told me they will better document this in the next version) if you mix in the other way around your results will most probably be WRONG !
Solid physics has a type of ALE with the spatial frame approach, therefore the Solid and ALE interact, normally one do not apply ALE to the solid domain, could this be your case, then disable the ALE domains in the solid physics node

Perhaps a screen sshot could help us better understand

--
Good luck
Ivar

Hi,

Thanks. It works now. I somehow disabled the free deformation that is was in shadow....

Best,
Stacey

Hi

what I read there is that the ALE (applied by default to all, is disabled for 1&3 becausethey have been included into the solid, but domain 2 the air/vacuum in the middle should not be modelled as a solid and is now modelled by the ALE to have its mesh follow the top and bottom plate deformations.

In your version the default physics nodes are not flagged with a "D" in the upper left corner of their respective Icons, that must appear then in v4.2A, this makes it easier to identify the minimum default sub-nodes.

Then I see you have disabled the two Feee deformation (domin) sub-node (that should point to "2") and the Prescribed (Boundary) Displacement that should point to the two horizotal line interfaces solid - air, with u,v as X,Y respective displacement.
The two first sub-nodes shoul be left as default, with X=0 and Y=0 as prescribed displacement.

--
Good luck
Ivar

Hi, Ivar,

Thanks for the help.

Now I got another question. When I am trying to add a "parametric sweep" to "study 1", I am now so confused about how to add the parametric sweep and how COMSOL assigns step number.

1) If there are several parameters I want to sweep, I should create several different "parametric sweep", right? Otherwise they will share the same parametric value?

2)Sometimes when I right click "Study 1" and choose "show default solver", the "Job Configurations" will also be generated, but sometimes it won't be generated. Since in the "Job Configurations", but there is some parametric step in the solver sublist, I see that we can define the parametric to run which solver,

I attached the captures.

Thanks.

Best,
Stacey

Hi

Be aware there are the two "parametric sweeps" in COMSOL. I often call them inner and outer loops.

The inner loop is defined in certain solvers as the Extension - Continuation sweep. This means for a given geometry and mesh, once the solver has solved for parameter value 1, it continues for parametervalue 2 (using the initial conditions of parameter value 1 results), etc.
You may add several parameters in the same go, but sorted out (#1.1 #1.2 ...) (#2.1 #2.2 ...) ... the () are only to show which are taken for the first set, then the second etc
You will get one data set that is a combination of all parameter values
check the doc on "solver continuation"

The outer loop is the specific "Parameter sweep" node you are using in your pdf example. This loops all the way around the geometry and the mesh, so you can change the geometry dimension (provided you do not change the topology such that COMSOL renumbers the Entity id's and mixes up the BC and domains !!)
The Parametric Sweep solver ALWAYS restart at "0" or your initial conditions, it does norally NOT use the results from previous parameter value solved.
The parameter (there can be several), obey the same rule as the "continuation sweep", but if you want to do a complete sweep for all values, using several nodes as you do is the correct way (you will get many stored values though;)
check the doc on "Parametric sweep"

Normally everything is written in the doc & help, at least in my 4.2a, but its somewhat scattered, so use an indexer on the pdf files, and use your dynamic help systematically, even when you do believe you know the stuff, I'm going on that way, still ;)

General requirement, the FIRST parameter (of any type continuation or sweep) MUST be monotone (increasing or decreasing), but the first parameter might be a dummy one, not used in any physics, just there to identify the sequence. But this makes the derived values table extracting sometimes tricky.
More, the parameter must be REAL and has NO units

Hope it helps, but its only my summary of reading through the doc just now ;)

For the solver nodes I leave mostly COMSOl take care, indeed with a continuation or single Parameteric sweep node, it adds a parameter controller sub node, and then if several paramters are there an additional job sequencing. For me it works mostly OK, One only ned to take some care in the Study list, as often you have 2 studies, one for the last iteration and one for the global. Now I' m not sure if a multi Paraemtric sweep is not adding several more study steps, to be tested

--
Good luck
Ivar

Hello Iva,
My aim is to model a strain gauge, which change their resistance by applied strain. But I saw this interesting contribution, which sound very seminar to my problem. So I try to follow your instruction and modelled the changing capacitance causes by the moving plates. But I got only the results from solid mechanics simulation, but the capacitance and the electrical potential remain the same. (I am using version 4.3)
Can you post a working .mph-file of this model or show me my mistake in the attached file please.
Thanks
Nghia

Hello Yemin Tang

Can you send me the file of simulating two physics sequentially: the first one is solid mechanics and the second one is electrostatic. My geometry is simply two parallel plates and wants to sweep of the force and get the capacitance reading. under this question how to use deformed mesh ?

Can you send or attach the working file for this case?

Regards
Islam

Dear all,

You can use the deformed geometry from structural mechanics as an input to another physics without necessarily using moving mesh or deformed geometry interfaces. Please see details in the following blog post.

www.comsol.com/blogs/how-to-reuse-a-deformed-shape-as-a-geometry-input/

Best Regards,
Temesgen Kindo, PhD
COMSOL Technical Support

Dear Dr.Temesgen Kindo

Thank you fro your reply

But now I have a problem with using Deformed Configuration , this message appears when using this option

Some geometric entities are missing.
- Vertex: 1, 2, 3, 4, 5, 6, 7, 8
See 'Problematic Deformed Mesh 1' to inspect the problematic mesh

Best Regards
Islam Hassan

Dear Islam

Please contact the support team for help with specific model files. Someone needs to look at your model file to figure out what is going on. Please send an email to support@comsol.com.

Best Regards,
Temesgen Kindo, PhD
COMSOL Technical Support

Hi,
Did you solve your problem? I am getting the same error when I want to use deformed configuration for next study step.

Thank you,
Morvarid