Seite 1 von 1

Plate and Shell elements

Verfasst: Di 21. Apr 2015, 19:52
von mturcios
Hello,

First off, thanks very much for the new version of Z88Aurora. I am very intrigued by the possibility of non-linear analysis. I was trying to start using shell and plate elements and was converting a mesh I created in Salome. At first I was having issues with the ordering of points as the face normal convention is different, but I got it imported properly (I believe). However, the results of my computation are incorrect. I noticed that the super mesh analysis says the determinants are incorrect, but I am not sure if this is an issue with the conversion, the original mesh, or my lack of familiarity with plate/shell elements.

I'm uploading the content of the z88i.txt file I generate as it is not too large. If anyone could help enlighten me that would be appreciated. Thanks!

The Salome element 208 is equivalent for the plate 20 or shell 23 (if nodes are in same plane)

z88 Mesh: https://www.dropbox.com/s/ll92por8495lp ... h.dat?dl=0
Salome Mesh: https://www.dropbox.com/s/ll92por8495lp ... h.dat?dl=0

Re: Plate and Shell elements

Verfasst: Mi 22. Apr 2015, 08:19
von SHautsch
Hello mturcios,

unfortunately your links only point to one single file: spearMesh.dat
I downloaded this file and had a quick look at it. Here are some general hints:

The structure seems OK for a simple conversion, the node list is pretty much the same as in Z88Aurora, only the number of DOF per node is missing.
The list of elements which is attached after the node list seems OK, too. Again only a simple change has to be made: add a line break and adapt the element number. Then add the correct first line to your file and you got yourself a working Z88 structure.

I also see a problem: your mesh contains two different elements: No. 103 and No. 208. Unfortunately it is not possible to calculate different element types in one part with Z88Aurora! You need to mesh your part with only one element type. The conversion then could be made with a simple perl script or something similar.

Please provide us with your z88i1.txt so we can have a detailed look at your problem.

Kind regards,
SHautsch

Re: Plate and Shell elements

Verfasst: Mi 22. Apr 2015, 21:59
von mturcios
So sorry, I pasted the same link twice. Here is the generated Z88 file:

https://www.dropbox.com/s/cu804k9feb0pymh/z88i.txt?dl=0

I have a small C program (its what I know how to use) program to only keep the element types I want (as this file format exports lines and such). The conventions are those of Code-Aster:

http://www.code-aster.org/outils/med/ht ... vites.html

The element normals are the opposite of Z88's, but I was able to take that into account upon conversion. I was able to get a decent looking mesh (right-side out elements), but when I tried to do a simple cantilever case, I
obtain very strange results. Let me know if the mesh is any good, and if so where else my problem could be.

The case I want to try is here:

https://www.dropbox.com/s/nnwgecrtw2hgb ... t.zip?dl=0

Thanks!

Re: Plate and Shell elements

Verfasst: Do 23. Apr 2015, 08:09
von SHautsch
Hello mturcios,

no problem. Now we've got your project and structure file and were able to check it.

First of all your structure (z88i1.txt) file with plate elements seems not to work out of the box because the element numbering is wrong. I guess that it's a general problem with conversion...you will have to check your C program. The correct numbering is displayed in the theory manual or element library.

And in my opinion the use of plate elements No.20 (which should only be loaded perpendicular to their surface) is not advisable. You should use shell elements like No. 23 or No. 24. If you would use elements No.23/24, you could just import a surface model via STEP or STL to Z88Aurora V3 and use our STL to shell converter to create your structure.
But you already got that right in your shellTest-Project.

Regarding the results of your project:
Your thickness seems quite small with 0.05 mm. Is that correct? That could explain strange displacement or stress values.
The main problem is that you calculated your structure with one of our iterative solvers, which tend to solve the structure even if there are some errors in it. If you try to run PARDISO on your project, you will get an error.
This is because your structure file is still wrong. Element No. 23 has six DOF per node, you only got 3 written in your structure file. This would be correct:

Code: Alles auswählen

3 3484 1144 20904 0 #AURORA_V3 
1 6 +1.000000E+001 +0.000000E+000 +0.000000E+000
... and so on for all nodes
And don't forget to set correct boundary conditions for these DOF (fix all 6 DOF, not only x/y/z as in your project).

The correct project with working structure and boundary conditions is attached.

Kind regards,
SHautsch

Re: Plate and Shell elements

Verfasst: Fr 24. Apr 2015, 00:27
von mturcios
Thank you so much! I am new to shell elements and hadn't thought about the degrees of freedom. I was able to fix my program and get the import/conversion working. Thanks for the help.

Re: Plate and Shell elements

Verfasst: Fr 24. Apr 2015, 06:50
von SHautsch
Dear mturcios,

you are welcome! Don't hesitate to ask another question if you experience problems with Z88Aurora :wink:

Kind regards,
SHautsch

Re: Plate and Shell elements

Verfasst: Mo 27. Apr 2015, 13:06
von TAH1712
Hello SHautsch.

I have only been reading this transcript, on behalf of this (myself) FEA learner driver ( and perhaps others to ), i would like to vote you a pay increase!

Kind regards..

Re: Plate and Shell elements

Verfasst: Di 28. Apr 2015, 07:37
von SHautsch
Hello TAH1712,

thank you very much :mrgreen: I'll talk to my boss, but since I'm in public service a pay increase only comes with age of service :lol:
But our team and I will surely keep up our work and give the best support we can!

Kind regards,
SHautsch