Das Forum zu Z88OS (Open Source) / The newsgroup for Z88OS (open source)

Moderatoren: ccad, mz15, auroraIco, Lehrstuhl

Benutzeravatar
vom finzi
#1896
Hallo :D
I would like to understand if I can make mixed mesh.
I am mainly interested in (shell or plate)+(solid) 3D meshes.
I couldn't find this information in the documentation (for all elements).
may be a smple matrix like this (for 3D analysis):
Code: Alles auswählen
			el_ty	01	10	17	16	23	24	04	02
hexa_8		01	YY	NN	YY	NN	NN	NN	YY	??
hexa20		10	NN	YY	NN	YY	??	??	YY	??
tetr_4		17	YY	NN	YY	NN	NN	NN	YY	??
tetr10		16	NN	YY	NN	YY	??	??	YY	??
quad_8		23	NN	??	NN	??	YY	YY	??	??
tria_6		24	NN	??	NN	??	YY	YY	??	??
trus_2		04	YY	YY	YY	YY	??	??	YY	??
beam_2		02	??	??	??	??	??	??	??	YY
may be that only displacements are calculate correctly and stress would be not?

Thanks for your support,
Andrea

PS I'm not sure the table is correct, it is just my understanding. :oops:
vom ccad
#1899
Hello Andrea,


your idea with the table is pretty good but let me give some deeper going explanations:

Basically, you may mix elements which feature the same type of dof. The number of dof may vary, however. As a matter of fact, this is true for most of the commercial FEA programs (sometimes they use so-called link elements for combining different element types but I think that's a no-good idea). Thus you may mix in Z88Aurora and Z88V14:

- Hexahedrons No.1&10 with Tetrahedrons No.16&17 with Trusses No.4 with Volume Shells No.21 & No.22 because every node has 3 dof ux, uy and uz. Beam No.2 should work, too, because the first 3 dof are in line with the dof of tetrahedrons, hexahedrons and trusses. Please note to define the entry "6" for nodes which are connected to beams no. 2 and, say, hexahedrons no.10 --> the higher number of dof counts. And be sure to give the correct (i.e. enough) boundary conditions for such nodes. A typical case is, to use trusses no.4 as "spring" or elastic boundary conditions for the volume elements.

- Plane Stress Elements No. 7 with No. 11 with No. 14 with No.3 with Truss No.9 because every node has 2 dof ux and uy. For beams no.13, see above.

- Torus Elements No.8 with No.12 with No.15 with No.6 because every node has 2 dof uradius and uheight. Thus, they cannot combined with other element types.

- Plate Elements No.18 with No.19 with No.19. You cannot combine plates (dof: uz, thetax, thetay) with, say, hexahedrons and tertahedrons (dof: ux, uy, uz). And this would make no sense in any way: plates are 2D elements (they are geometrically only defined by x and y coordinates but the first defection goes into z direction) while tetrahedrons and hexahedrons are really 3D elements (they are geometrically defined by x, y, and z coordinates).

- Shell elements No.23 & No.24. Basically, you could combine volume elements (dof: ux, uy, uz) with, say, hexahedrons and tetrahedrons (dof: ux, uy, uz and thetax, thetay, thetaz). But I would not recommend this: Try to imagine this case geometrically, i.e. how to connect the elements in reality. And because thetax and thetay are global dof it would be rather difficult ot apply correct boundary conditions. Thetaz is quite useless; it's a so-called pseudo (or zombie) dof.

Best Regards

Prof. Rieg
Benutzeravatar
vom finzi
#1906
Thank Prof. Rieg for your detailed answer.

More in general I would like to thank you for all effort you do to reply to all questions.
And I want thank you because in 30 year "playing" with FE, this is the first time I can go so deep in the coding.
Your Z88 is really a great present for me. THANK!!!!!

Then, I understood the problem and I tried to prepare my element compatibility table:
green is OK, yellow means "check mesh continuity", red means "possible but not recomended" different numer of dof.

Best regards,
Andrea
Dateianhänge
Elements_compatibility_table1.jpg
xls file available
Elements_compatibility_table1.jpg (211.75 KiB) 2491 mal betrachtet
vom ccad
#1913
Dear Andrea,

thank you for your nice words amd your table! I've checked your table roughly and it seems okay to me (but please let my co-workers do a precise check .. we'll respond in some days). Anyway, your table is a very good idea and i think we'll present such a table in our future manuals.

Regards

Prof. Rieg
Start

Hallo Schlorchi00, leider konnten wir den Fehler[…]

Hallo liebes Z88 Team, gibt es bereits Neuigkeite[…]

UPDATE Nachdem ich es nach weiteren Tests leide[…]

We're glad you got it working. We will have a look[…]