Hiii,
I have got one findings...To check the problem I executed same file with stress solver,,, then it's showing damping effect...But when using elastic solver it's not generating damping effect...
Damping problem
-
- Posts: 2237
- Joined: 25 Jan 2019, 01:28
- Antispam: Yes
Re: Damping problem
When you said did I generate my own mesh, I did translate it from a gmsh file you posted. When I translate gmsh I use
ElmerGrid 14 2 meshname.msh -autoclean
the auto clean flag eliminates the 1 node and 2 node boundary elements from a 3D mesh. I used
ElmerGrid 14 2 meshname.msh
without the autoclean and the 1 node 2 node boundary elements stay in, this should not have been a problem, but when I left them in I received a poor solution. try
ElmerGrid 14 2 meshname.msh -autoclean
and see if it fixes the issue. you should only have 303 and 504 elements.
ElmerGrid 14 2 meshname.msh -autoclean
the auto clean flag eliminates the 1 node and 2 node boundary elements from a 3D mesh. I used
ElmerGrid 14 2 meshname.msh
without the autoclean and the 1 node 2 node boundary elements stay in, this should not have been a problem, but when I left them in I received a poor solution. try
ElmerGrid 14 2 meshname.msh -autoclean
and see if it fixes the issue. you should only have 303 and 504 elements.
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Damping problem
Hi
This is really annoying problem. I looked a little on the root cause. In Gmsh 4.1 format, see
https://gmsh.info/doc/texinfo/gmsh.html#MSH-file-format
Now it seems that "entitytag" gets the same value for different dimensionalities. For example, here we have for surface elements 1..6, for line elements 1..12, and for point elements 1..8. Like here we have BCs set for 3 and 4, that means that they will also be set for some random selection of line and point elements as well - if they are not removed from the set.
I don't know whether this is something that has changed over different versions. It seems that detdp separates the elements by their dimension and thereby you can use same tags. Elmer only separates bulk elements vs. boundary elements so you cannot use the same tags.
I will do some changes to resolve this. Probably checking that lower dimensional boundary elements have some offset.
With the current code the "-autoclean" flag obviously gets rid of the problem as does use of physical entities that are number uniquely in Gmsh.
-Peter
This is really annoying problem. I looked a little on the root cause. In Gmsh 4.1 format, see
https://gmsh.info/doc/texinfo/gmsh.html#MSH-file-format
Code: Select all
$Elements
numEntityBlocks(size_t) numElements(size_t)
minElementTag(size_t) maxElementTag(size_t)
entityDim(int) entityTag(int) elementType(int; see below)
numElementsInBlock(size_t)
elementTag(size_t) nodeTag(size_t) ...
...
...
I don't know whether this is something that has changed over different versions. It seems that detdp separates the elements by their dimension and thereby you can use same tags. Elmer only separates bulk elements vs. boundary elements so you cannot use the same tags.
I will do some changes to resolve this. Probably checking that lower dimensional boundary elements have some offset.
With the current code the "-autoclean" flag obviously gets rid of the problem as does use of physical entities that are number uniquely in Gmsh.
-Peter
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Damping problem
Ok, there should be a fix now in "devel". This is pretty safe one because it is very seldom that the user wants to use the same tag for 2D and 1D boundary conditions, for example.
It seems that the original .msh files has some physical entities set for the boundaries that we also want to use. However, all entities are saved and there is a warning about it. Saving just the physical entities would have circumvented this. Now ElmerGrid has to interpret what to do with entities where for some the physical entity has been set and for many not. You see this as warnings. The new operation should make the lower dimensional entities harmless even if they are not eliminated.
-Peter
It seems that the original .msh files has some physical entities set for the boundaries that we also want to use. However, all entities are saved and there is a warning about it. Saving just the physical entities would have circumvented this. Now ElmerGrid has to interpret what to do with entities where for some the physical entity has been set and for many not. You see this as warnings. The new operation should make the lower dimensional entities harmless even if they are not eliminated.
-Peter
-
- Posts: 2237
- Joined: 25 Jan 2019, 01:28
- Antispam: Yes
Re: Damping problem
Thanks Peter,
It tested fine and the lower dimensional entities are harmless if the tags are different for the boundary elements.
Kevin
It tested fine and the lower dimensional entities are harmless if the tags are different for the boundary elements.
Kevin
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Damping problem
Thank you Kevin!
Re: Damping problem
Hiiiii,
For that whether after executing that autoclean line for my mesh in command prompt, I have to incorporate same .msh file from ElmerGUI?
For that whether after executing that autoclean line for my mesh in command prompt, I have to incorporate same .msh file from ElmerGUI?
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Damping problem
Hi,
Funnily ElmerGUI does the "-autoclean" by default as it cannot deal with the lower dimensional entities anyway. So this only affected the process of reading .msh files on command line.
-Peter
Funnily ElmerGUI does the "-autoclean" by default as it cannot deal with the lower dimensional entities anyway. So this only affected the process of reading .msh files on command line.
-Peter
Re: Damping problem
Hiii,
Thanks peter..If elmergui does this autoclean by default, then to fix the issue of not getting any effect of damping in case of elasticsolver what should I do next? I am not getting any way to get out of it..
Thanks peter..If elmergui does this autoclean by default, then to fix the issue of not getting any effect of damping in case of elasticsolver what should I do next? I am not getting any way to get out of it..
-
- Posts: 2237
- Joined: 25 Jan 2019, 01:28
- Antispam: Yes
Re: Damping problem
You said your mesh.header and mesh.boundary was different than mine, can you post them so I can see the difference?