Elastic Plates vs Linear Elasticity

General discussion about Elmer
stoykov
Posts: 26
Joined: 11 May 2012, 13:18
Antispam: Yes

Elastic Plates vs Linear Elasticity

Post by stoykov » 14 May 2012, 12:05

Hello,

I just started to use Elmer, and the first thing to do is to solve linear eigenvalue problem of elastic beams. I wish to compare the eigenvalues of thin beam which I obtained by Bernoulli-Euler or Timoshenko beam theories with Elmer. The problem is that if I use define the equation as Elastic Plates and define the beam as plate, with very small width, I obtain correctly the eigenvalues. But if I define the equation by using Linear Elasticity and the beam mesh is 3D, then I have much higher and obviously wrong eigenvalues. Does anyone have an idea why the problem occurs for Linear Elasticity models?

Thanks,
Stan

raback
Site Admin
Posts: 3240
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Elastic Plates vs Linear Elasticity

Post by raback » 14 May 2012, 13:13

Hi

I would guess that you might be using linear elements and are experiencing "locking". Try setting in Solver section

Code: Select all

Element = p:2
To use 2nd order elements.

-Peter

stoykov
Posts: 26
Joined: 11 May 2012, 13:18
Antispam: Yes

Re: Elastic Plates vs Linear Elasticity

Post by stoykov » 14 May 2012, 14:57

Thank you Peter, now it works very good :)

Could you please tell how can I define higher order elements using ElmerGUI or do I have to define it in Gmsh (I am using Gmsh to define the mesh)?

Thank you in advance,
Stan

raback
Site Admin
Posts: 3240
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Elastic Plates vs Linear Elasticity

Post by raback » 14 May 2012, 15:40

Hi

You can do that in Gsmh by choosing quadratic elements. In ElmerGUI you can probably set the "Element = p:2" in free text input of the Solver specific options. This creates p-elements which are however, not the same as the Lagrange elements created by Gmsh. The p-elements are hierarchical which may result to easier matrix equations. Unfortunately the p-elements cannot increase the accuracy of the geometric presentation. If this is your wish the quadratic elements in Gmsh may be better choice.

-Peter

sharath
Posts: 15
Joined: 14 Jul 2017, 09:20
Antispam: Yes

Re: Elastic Plates vs Linear Elasticity

Post by sharath » 20 Jul 2017, 09:29

I have a question on the above topic. I am trying to simulate a axial bar problem with 1x1x1 mts dimensions. i am able to import the linear elements from Gmsh to ElmerGUI and its giving me pity much nice answerers. If i am using Gmsh created the quadratic elements, ElmerGUI is recognising only on boundary and one surface element. I am unable to chose other surfaces for applying boundary conditions.
gmshelements.png
(160.33 KiB) Not downloaded yet
The above fig is the mesh created in the Gmsh. But,

Elmer.png
(126.83 KiB) Not downloaded yet
I am unable to get the 2nd order elements to be recognised by ElmerGUI as i have shon in the above figure.

Can any one help me ...

mzenker
Posts: 1729
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Elastic Plates vs Linear Elasticity

Post by mzenker » 20 Jul 2017, 10:59

Hi,

this may be related to physical groups in gmsh. You may open your mesh file with gmsh to see what was actually saved. You have to declare every surface/volume you want to be saved as physical, or you have to check "Save all (ignore physical groups)" when saving the mesh with gmsh.

HTH,

Matthias

sharath
Posts: 15
Joined: 14 Jul 2017, 09:20
Antispam: Yes

Re: Elastic Plates vs Linear Elasticity

Post by sharath » 20 Jul 2017, 13:50

Thank you for your quick response. On many iterations, i have found that ElmerGUI may not able to detect type of Elements. Because ElmerGUI has a very few verity of Element Library. If I am using unstructured grid all the surfaces are been detected by the ElmerGUI and its taking Higher order elements also like(Tetrahedron).

For structured grid, i have rechecked the declaration of all the surfaces and volumes. They are fine in Gmsh but once we up load into Elmer its detecting only one surface.


Thanking you,

Sharath.

mzenker
Posts: 1729
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Elastic Plates vs Linear Elasticity

Post by mzenker » 20 Jul 2017, 14:59

Hi,

please try the following:

* Open within gmsh the exact mesh file which you want to read with Elmer. Are all entities there?
* If yes: open the same mesh file with ElmerGUI. Are all entities there now?
* If still not all entities are displayed, then maybe the entities are detected, but not displayed. Make sure that "View->Surface Mesh" and "View->Volume Mesh" are checked (volume mesh display may slow down the GUI). Examine "Model->Summary". Is the number of detected entities correct? You can also examine the checkboxes you get under "Model->Boundary Condition" and "Model->Body Force".
* If the GUI doesn't detect your entities, you may try with the ElmerGrid commandline version using the autoclean option (see ElmerGrid manual) and proceed without using the GUI. The ElmerGrid output may give you an indication if all the entities are there.

HTH,

Matthias

sharath
Posts: 15
Joined: 14 Jul 2017, 09:20
Antispam: Yes

Re: Elastic Plates vs Linear Elasticity

Post by sharath » 20 Jul 2017, 20:04

Yes I have opened the Exact file in Gmsh. It was fine.
Gmshfiles.zip
(2.9 KiB) Downloaded 28 times
These are the .geo and .msh files that i have created in Gmsh. Once .msh file is imported into ElmerGUI, following is the Summary

******************************************************** Elmer Import Summary
FINITE ELEMENT MESH
Mesh dimension: 3
Leading element dimension: 3
Nodes: 189
Volume elements: 10
Surface elements: 1
Edge elements: 4
Point elements: 0

ELEMENT TYPES
827: 10
409: 1
203: 4

BOUNDING BOX
X-coordinate: [ 0 , 1 ]
Y-coordinate: [ 0 , 1 ]
Z-coordinate: [ 0 , 1 ]

GENERAL
Equations: 0
Materials: 0
Boundary conditions: 0
Body properties: 0

VOLUME BODIES
Body 1: 10 volume elements
Undetermined: 0
Total: 1 volume bodies

SURFACE BODIES
Undetermined: 0
Total: 0 surface bodies

EDGE BODIES
Undetermined: 0
Total: 0 edge bodies

SURFACE BOUNDARIES
Undetermined: 0
Total: 0 surface boundaries

EDGE BOUNDARIES
Undetermined: 0
Total: 0 edge boundaries
**********************************************************************
and these are the Elmer Project files. My problem is that Elmer is not detecting the edge and boundary surface elements.

I have also checked all the entities. They are showing that it has recognised ten volume elements and one body.
I have also tried with ElmerGrid. It has created mesh(header,Boundary,Elements,nodes) file which are also same as ElmerGUI files.
I am attaching the project file below. Please help me if any body can..
Gmshfiles.zip
(2.9 KiB) Downloaded 28 times

mzenker
Posts: 1729
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: Elastic Plates vs Linear Elasticity

Post by mzenker » 21 Jul 2017, 11:00

Hi,

I have had a look at your files. For me also, Elmer doesn't recognize the surfaces except one. So I opened your goe file in gmsh, meshed, saved, and Elmer read the file just OK.
Two methods worked for me:

1. Just mesh and save, checking "Save all (ignore physical groups)" (cube_all.msh)
2. Mark all surfaces and the volume as physical and save (cube_physical.msh)

I don't know why it didn't work for you...

HTH,

Matthias
Attachments
cube_physical.msh
(1.96 KiB) Downloaded 22 times
cube_all.msh
(2.9 KiB) Downloaded 22 times

Post Reply