Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

3D Stress>Boundary Conditions>Flat Boundary Under Stress [Kinematic Constraints]

Please login with a confirmed email address before reporting spam

Hello,

I'm trying to model Biaxial compression on a thin sheet with dimensions:

X = 3 microns
Y = 5 microns
Z = 0.34 nanometers

with normal Stresses (biaxial) inward:

sigma_x1 = 0.05 TPa (left boundary)
sigma_x2 = -0.05 TPa (right boundary)
sigma_y1 = 0.05 TPa (bottom boundary)
sigma_y2 = -0.05 TPa (top boundary)

And I need to specify the following boundary constraints:

All boundaries are free to deform except left and right boundaries completely rigid, or flat.

That is to say, the Y by Z faces on the left and right side need to be kept flat, or rigid, under stress.
---
I found in the Structural Mechanics Model User's Guide (COMSOL 3.5a) on page 83 under "Kinematic Constraints," an example called "Straight Edge Constraint by Equations" explains how to keep an edge straight under stress.

However, this example uses complicated Matrix equations and only deals with a straight edge, whereas I need to keep an enitre boundary, or face, straight/flat.
---
Is there some simple boundary setting parameter that I can specify to do this in a 3D module?

Should I be directed to an example model to illustrate this?

Which application module should I use to do this [I tried 3D Solid, Stress-Strain (smsld) with no success].

I've attached the model file I've been working on.

Thanks for the help,
Kevin


7 Replies Last Post 2010年8月4日 GMT-4 02:19
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年8月1日 GMT-4 04:39
Hi

with such an aspect ratio, do you not believe you could just as well use the shell elements ? thenyou would be closer to the example too, no ?

--
Good luck
Ivar
Hi with such an aspect ratio, do you not believe you could just as well use the shell elements ? thenyou would be closer to the example too, no ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年8月2日 GMT-4 23:51
Thanks for responding Ivar,

Although you are right that my aspect ratio is quite large (~10,000), however, the main concern is to establish a flat/rigid surface boundary. Will I be able to do this with the shell elements, that is, with the Shell Application Mode?

It seems the only available boundary conditions in the Shell Mode [Version 3.5a] are: free, fixed, pinned, no rotation, prescribed displacement, and symmetry planes, none of which can specify a flat, rigid boundary under compression.

Will I instead need to specify some predetermined constants and matrices? If so, any advice on how to do this? Or, are there any pre-existing models that you may know of which may either have flat, rigid boundaries under stress or similar matrices which may help in my case?

Thanks again.

Best regards,
Kevin
Thanks for responding Ivar, Although you are right that my aspect ratio is quite large (~10,000), however, the main concern is to establish a flat/rigid surface boundary. Will I be able to do this with the shell elements, that is, with the Shell Application Mode? It seems the only available boundary conditions in the Shell Mode [Version 3.5a] are: free, fixed, pinned, no rotation, prescribed displacement, and symmetry planes, none of which can specify a flat, rigid boundary under compression. Will I instead need to specify some predetermined constants and matrices? If so, any advice on how to do this? Or, are there any pre-existing models that you may know of which may either have flat, rigid boundaries under stress or similar matrices which may help in my case? Thanks again. Best regards, Kevin

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年8月3日 GMT-4 01:18
Hi

if you are in shell mode, your side surface is only an edge and you can apply the theory from the doc directly, no ?

You can also attach shells with very high E along the sides of your 3D model, it's not very "good physics" but could give you some results (to be validated carefully though)

--
Good luck
Ivar
Hi if you are in shell mode, your side surface is only an edge and you can apply the theory from the doc directly, no ? You can also attach shells with very high E along the sides of your 3D model, it's not very "good physics" but could give you some results (to be validated carefully though) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年8月3日 GMT-4 02:12
Hello,

I appreciate the advice, however, it seems I've realized two problems with the Shell Mode:

1) I do not wish to model a shell, rather, I wish to model a continuous solid. This is apparent in that the Shell Physics settings has no Subdomain setting, when indeed I need to specify a Young's Modulus in the subdomain.

2) For my rather large aspect ratio of about 10,000, the Mesh compiler (which can only Mesh triangular elements in Shell Mode) has great difficulty generating a triangular Shell mesh, even when using Extremely Coarse Mesh sizes.

I like the method you've proposed about specifying straight edges directly from the aforementioned document's theory and simply specifying an extremely large Young's Modulus on those sides, however, I do not believe I can accomplish this with my given aspect ratio and Mesh generation failure.

Any suggestions? I cannot decrease my aspect ratio as that will render my model useless. In fact, even after decreasing the aspect ratio by a factor of 10, the Mesh still cannot be generated with an Extremely coarse mesh size anyways.

Thanks again for your help.

Best,
Kevin
Hello, I appreciate the advice, however, it seems I've realized two problems with the Shell Mode: 1) I do not wish to model a shell, rather, I wish to model a continuous solid. This is apparent in that the Shell Physics settings has no Subdomain setting, when indeed I need to specify a Young's Modulus in the subdomain. 2) For my rather large aspect ratio of about 10,000, the Mesh compiler (which can only Mesh triangular elements in Shell Mode) has great difficulty generating a triangular Shell mesh, even when using Extremely Coarse Mesh sizes. I like the method you've proposed about specifying straight edges directly from the aforementioned document's theory and simply specifying an extremely large Young's Modulus on those sides, however, I do not believe I can accomplish this with my given aspect ratio and Mesh generation failure. Any suggestions? I cannot decrease my aspect ratio as that will render my model useless. In fact, even after decreasing the aspect ratio by a factor of 10, the Mesh still cannot be generated with an Extremely coarse mesh size anyways. Thanks again for your help. Best, Kevin

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年8月3日 GMT-4 03:49
Hi

I do not fully follow you, among others because:

1) in you model you have only 1 element in the height/thickness you should at least have 3, ideally many more. your mesh proportions like that are horribles and prone to severe numerical errors.
2) you use symemtry conditions (yz) on two sides, this is equivalent to state that lcoal u=0 no displacement, + normal pressure, is this what you want ?
3) if you use a shell you need only a surface, not a "volume" and the thicknes is simulated by the equations (anyhow much better than a 1 element thick 3D brick)

Finally I do still not catch exactly what you want to fix (without overconstraining your model), so I have some difficulties to propose something further

--
Good luck
Ivar
Hi I do not fully follow you, among others because: 1) in you model you have only 1 element in the height/thickness you should at least have 3, ideally many more. your mesh proportions like that are horribles and prone to severe numerical errors. 2) you use symemtry conditions (yz) on two sides, this is equivalent to state that lcoal u=0 no displacement, + normal pressure, is this what you want ? 3) if you use a shell you need only a surface, not a "volume" and the thicknes is simulated by the equations (anyhow much better than a 1 element thick 3D brick) Finally I do still not catch exactly what you want to fix (without overconstraining your model), so I have some difficulties to propose something further -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年8月3日 GMT-4 20:29
Thanks again for responding, Ivar.

I'm sorry for causing confusion, but now it seems that I'm confused.

Replying from concerns 1-3:

1) What do you mean I only have one element in height/thickness when I should have three? i.e., from my dimesions:

X = 3 microns,
Y = 5 microns,
Z = 0.34 nanometers,

You're saying that my Z-dimension is the single element, when I should have 3? How would I implement 3 and how would that help my mesh generate?

2) I wish for a surface (with thickness as mentioned to be specified in Shell Mode settings) with the X sides to be completely free to deform, while the Y sides to be flat/rigid, however, I wish for all 4 sides to be free to displace. So my two Y sides will be rigid and will be displaced inward during biaxial compression, so there will indeed be a u displacement, as well as v and w displacements.

3) Regarding creating a surface (not a volume) in the Shell application, I've tried creating a plane in the Shell application, but the Structural Mechanics>Shell application is only available in 3D, correct? So when I attempt to create a plane: Draw>Work Plane Settings>x-y plane, I am left with a simple plane as desired, however, I then cannot access Subdomain Settings, Boundary settings or Edge Settings under the Physics tab. Essentially, it seems that I cannot apply any physics settings to a simple 2D plane in this mode. It seems from here that I must extrude to 3D, but that is just the same as creating a 3D cube-type shape in the first place.

So I understand my goal is to create a plane in the Shell Mode and then my edges become boundaries, in which I can then implement the straight edge boundary condition from the documentation mentioned, however, I cannot seem to create this model.

Is there a way to use Shell Mode in 2D? [for version 3.5a that is]

Or rather, am I supposed to create a 3D cube and then only apply settings to one face of the cube, applying proper edge settings and so forth? If so, this doesn't seem to work because after applying stress all 4 edges remain straight and non-displaced (because they're fixed to the cube) when I wish for two edges to remain straight and all 4 edges to displace under stress.

Thanks for the help.

Best regards,
Kevin
Thanks again for responding, Ivar. I'm sorry for causing confusion, but now it seems that I'm confused. Replying from concerns 1-3: 1) What do you mean I only have one element in height/thickness when I should have three? i.e., from my dimesions: X = 3 microns, Y = 5 microns, Z = 0.34 nanometers, You're saying that my Z-dimension is the single element, when I should have 3? How would I implement 3 and how would that help my mesh generate? 2) I wish for a surface (with thickness as mentioned to be specified in Shell Mode settings) with the X sides to be completely free to deform, while the Y sides to be flat/rigid, however, I wish for all 4 sides to be free to displace. So my two Y sides will be rigid and will be displaced inward during biaxial compression, so there will indeed be a u displacement, as well as v and w displacements. 3) Regarding creating a surface (not a volume) in the Shell application, I've tried creating a plane in the Shell application, but the Structural Mechanics>Shell application is only available in 3D, correct? So when I attempt to create a plane: Draw>Work Plane Settings>x-y plane, I am left with a simple plane as desired, however, I then cannot access Subdomain Settings, Boundary settings or Edge Settings under the Physics tab. Essentially, it seems that I cannot apply any physics settings to a simple 2D plane in this mode. It seems from here that I must extrude to 3D, but that is just the same as creating a 3D cube-type shape in the first place. So I understand my goal is to create a plane in the Shell Mode and then my edges become boundaries, in which I can then implement the straight edge boundary condition from the documentation mentioned, however, I cannot seem to create this model. Is there a way to use Shell Mode in 2D? [for version 3.5a that is] Or rather, am I supposed to create a 3D cube and then only apply settings to one face of the cube, applying proper edge settings and so forth? If so, this doesn't seem to work because after applying stress all 4 edges remain straight and non-displaced (because they're fixed to the cube) when I wish for two edges to remain straight and all 4 edges to displace under stress. Thanks for the help. Best regards, Kevin

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年8月4日 GMT-4 02:19
Hi

you have many questions there, as usual when working with COMSOL you need to become fluent in many subjects from advanced matths to multi-physics :)

For the meshing of solids, you have a special option to "distort" a direction (usually x,y,z) during the mesh process, this will internally scale the defined direction, then apply the mesh and then reduce the size by the inverse of the scale (just as in 3.5a in 3D you can double click on the "EQUAL" lower GUI frame note and when you hit "refresh view" you will get different graphical scales along x,y,z such that your thin plate becomes a "thick box" this is very useful for helping selecting the "small" items. Read through (a couple of times) the doc on meshing, (I would say that the 3.5 doc is still more detailed than the v4 doc, today)

You can also use the sweep mesh (on simple geometries) and say how many elements you want in the thickness, or you can also start meshing the edges, the surfaces and then the volumes (COMSOL has a nice 1 day course on advanced meshing techniques, worth the visit)

Now for shell elements, you can either work in 2D, or in 3D (I expect easier for your case the way I beleive I understand it) you create a thin feature (just a surface i.e. by setting a workplane drawing the square and then embedded it into the 3D (extrusion of "0" thickness) then you have your "3D shell surface" with edges defining the height, you must then set all the related parameters

Hope this helps on the way

By the way the bet learining curve with COMSOL is to take the courses with a few months in between, and do not start by mixing too many physics, one at the time is anyhow the only way to build up a complex model

--
Good luck
Ivar
Hi you have many questions there, as usual when working with COMSOL you need to become fluent in many subjects from advanced matths to multi-physics :) For the meshing of solids, you have a special option to "distort" a direction (usually x,y,z) during the mesh process, this will internally scale the defined direction, then apply the mesh and then reduce the size by the inverse of the scale (just as in 3.5a in 3D you can double click on the "EQUAL" lower GUI frame note and when you hit "refresh view" you will get different graphical scales along x,y,z such that your thin plate becomes a "thick box" this is very useful for helping selecting the "small" items. Read through (a couple of times) the doc on meshing, (I would say that the 3.5 doc is still more detailed than the v4 doc, today) You can also use the sweep mesh (on simple geometries) and say how many elements you want in the thickness, or you can also start meshing the edges, the surfaces and then the volumes (COMSOL has a nice 1 day course on advanced meshing techniques, worth the visit) Now for shell elements, you can either work in 2D, or in 3D (I expect easier for your case the way I beleive I understand it) you create a thin feature (just a surface i.e. by setting a workplane drawing the square and then embedded it into the 3D (extrusion of "0" thickness) then you have your "3D shell surface" with edges defining the height, you must then set all the related parameters Hope this helps on the way By the way the bet learining curve with COMSOL is to take the courses with a few months in between, and do not start by mixing too many physics, one at the time is anyhow the only way to build up a complex model -- Good luck Ivar

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.