Hexahedron super-element meshing - bad node numbering ?

Alle Fragen zu: Vernetzung, Materialien, Lasten, Randbedingungen und Elementparameter /
All questions to: meshing, materials, boundary conditions and element properties

Moderatoren: ccad, mz15, auroraIco, Lehrstuhl

Antworten
Siegfried
Newbie
Beiträge: 5
Registriert: Mi 26. Sep 2012, 08:57

Hexahedron super-element meshing - bad node numbering ?

Beitrag von Siegfried »

I am using Z88 Aurora V2 ( 09-Aug-2012 )for a simple model using hexahedron elements, type 1.

The structure is a flat plate, 500x500x20 mm, imported as a single hexahedron element in a Z88I1.TXT file:
3 8 1 24 0 0 0
1 3 0 0 0
2 3 500 0 0
3 3 500 500 0
4 3 0 500 0
5 3 0 0 20
6 3 500 0 20
7 3 500 500 20
8 3 0 500 20
1 1
1 2 3 4 5 6 7 8

In Z88Aurora this is meshed 8x8x1 with Type 1 hexahedron elements, using super-element meshing. After setting material ( 'Engineering Steel' ) and constraints ( - one narrow face fixed and a point load at one other corner - ), the solver crashes ( - any one of: SICCG, SORCG, PARADISO - ) and Z88R.LOG reports:
...
...
start Z88BR
### Element 1 ###
### Jacobi determinant zero or negative ..stop ###
### element numbering wrong, not counter-clockwise ###
### Recover:renumber wrong elements(consult manual)###

... but the elements have been automatically generated !

Any ideas ?
Benutzeravatar
MNeidnicht
Alumni
Beiträge: 47
Registriert: Do 2. Sep 2010, 16:01

Re: Hexahedron super-element meshing - bad node numbering ?

Beitrag von MNeidnicht »

Hello Siegfried,

the problem is that even your one-hexahedron-model has a wrong element numbering. This means, that all surface-normals show to the middle of the part and not outside, as they should. The colour of the part in this case is black instead of yellow. Please try this z88i1.txt-file:

3 8 1 24 0 0 0
1 3 0 0 0
2 3 500 0 0
3 3 500 500 0
4 3 0 500 0
5 3 0 0 20
6 3 500 0 20
7 3 500 500 20
8 3 0 500 20
1 1
5 6 7 8 1 2 3 4

Greetings
Martin Neidnicht
Siegfried
Newbie
Beiträge: 5
Registriert: Mi 26. Sep 2012, 08:57

Re: Hexahedron super-element meshing - bad node numbering ?

Beitrag von Siegfried »

Thanks Martin - that all worked well.

My next problem is when I add a second layer - Z88I1.TXT shown below:
3 12 2 36 0 0 0
1 3 0 0 0
2 3 500 0 0
3 3 500 500 0
4 3 0 500 0
5 3 0 0 20
6 3 500 0 20
7 3 500 500 20
8 3 0 500 20
9 3 0 0 40
10 3 500 0 40
11 3 500 500 40
12 3 0 500 40
1 1
5 6 7 8 1 2 3 4
2 1
9 10 11 12 5 6 7 8

This imports OK. I pick element 1 and use 'Add Set', then select this set for super-element meshing, to give a '10 E 10 E 1 E' mesh, but both super-elements get meshed.
Benutzeravatar
MNeidnicht
Alumni
Beiträge: 47
Registriert: Do 2. Sep 2010, 16:01

Re: Hexahedron super-element meshing - bad node numbering ?

Beitrag von MNeidnicht »

Yes. And this is exactly, what it should be. :wink:

What would happen if only the upper hexahedron would be divided? New nodes, which lie on the surface between the two elements then are no longer part of the lower hexahedron. Displacements can only be transmitted via node matchings. So the elements would be able to penetrate each other or the border could be opened up. To avoid this situation - so-called nonconforming meshes - the lower hexahedron is devided like the upper one.

Greetings
Martin Neidnicht
Siegfried
Newbie
Beiträge: 5
Registriert: Mi 26. Sep 2012, 08:57

Re: Hexahedron super-element meshing - bad node numbering ?

Beitrag von Siegfried »

OK. My example was a poor one. A better illustration would be if I wanted to mesh element 1:
10 E 10 E 1 E
and element 2:
10 E 10 E 2 E

Or the case in the attached image, which is a plate - the bottom of a box. I want to mesh the centre element 10x10, the edge elements 10x1 and 1x10, and leave the corners as single elements. These mesh-densities will then match the 4 sides of the box when they are added.
Dateianhänge
Plate example
Plate example
plate1.gif (3.18 KiB) 5031 mal betrachtet
Siegfried
Newbie
Beiträge: 5
Registriert: Mi 26. Sep 2012, 08:57

Re: Hexahedron super-element meshing - bad node numbering ?

Beitrag von Siegfried »

Z88I1.txt file for the above example is attached ( - zipped to make it upload. )

Is it possible to mesh this as described, 10x10x1 in the centre, 10x1x1 and 1x10x1 in the edges ?
Dateianhänge
Z88I1.zip
Plate super-element structure
(404 Bytes) 168-mal heruntergeladen
mifris
Alumni
Beiträge: 153
Registriert: Mo 27. Sep 2010, 20:41

Re: Hexahedron super-element meshing - bad node numbering ?

Beitrag von mifris »

Hello Siegfried,

it is not possible to mesh your example as you want to. But why don´t you mesh the main rectangle. Divide it like the same size of one "corner-element", i.d. 12 times in x-, 12 times in y- and one time in z-direction. Then you should have one part meshed, which you can assign different sets. :)

Best regards
M. Frisch
Siegfried
Newbie
Beiträge: 5
Registriert: Mi 26. Sep 2012, 08:57

Re: Hexahedron super-element meshing - bad node numbering ?

Beitrag von Siegfried »

OK. Thanks Michael.

( I think your method will work only if the plate size is an integer multiple of the wall-thickness ? )
mifris
Alumni
Beiträge: 153
Registriert: Mo 27. Sep 2010, 20:41

Re: Hexahedron super-element meshing - bad node numbering ?

Beitrag von mifris »

Your welcome. Yes your right, only an integer multiple is possible.

Have a nice day. :)
Antworten