Linear vs Quadratic mesh
- Joe Dowsett
- Topic Author
- Offline
- Junior Member
Less
More
- Posts: 22
- Thank you received: 0
14 years 3 months ago #4613
by Joe Dowsett
Linear vs Quadratic mesh was created by Joe Dowsett
Can someone briefly explain the circumstances in which either form of mesh should be used? And/or recommend a website / tutorial detailing considerations when creating meshes?
In even simple geometries I get quite different values for the stress occurring depending upon whether I use a quadratic or linear mesh. How should I decide which is the appropriate mesh to use?
In even simple geometries I get quite different values for the stress occurring depending upon whether I use a quadratic or linear mesh. How should I decide which is the appropriate mesh to use?
- Claus
- Offline
- Moderator
Less
More
- Posts: 670
- Thank you received: 34
14 years 3 months ago #4619
by Claus
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
Replied by Claus on topic Re:Linear vs Quadratic mesh
Im going out on a limp here, but for pretty much everything related to Aster, a quadratic mesh is preferred/required for precise results.
Someone one this forum recently mentioned that quadratic mesh isn't suited for contact simulation, but that is the only case I can think of were linear mesh is preferred over quadratic.
For reference:
Syrthes 3.4.2 will only accept quadratic meshes - (this will change in the upcoming version)
Code_Saturne will read quadratic meshes but convert them to linear meshes when starting the calculation.
Code_Aster will eat anything, but the results will heavily favour quadratic mesh*.
*When viewing e.g. VonMises stress in Salomé or where ever, it is important to postprocess the EQUI_ELGA_SIGM field and not the EQUI_NOUE_SIGM field; this is because the equiv_noue_sigm field is created by interpolating the results from equi_elga_sigm which can lead to imprecise results, especially where stress concentrations are high.
To further muddy the waters, more precise results from Aster are obtained by using quads/hexahedrons over triangles/tetrahedrons.
Kees has some more valuable information at:
www.caelinux.org/wiki/index.php/Contrib:...nt_to_solid_elements
Regards,
Claus<br /><br />Post edited by: Claus, at: 2010/08/29 15:17
Someone one this forum recently mentioned that quadratic mesh isn't suited for contact simulation, but that is the only case I can think of were linear mesh is preferred over quadratic.
For reference:
Syrthes 3.4.2 will only accept quadratic meshes - (this will change in the upcoming version)
Code_Saturne will read quadratic meshes but convert them to linear meshes when starting the calculation.
Code_Aster will eat anything, but the results will heavily favour quadratic mesh*.
*When viewing e.g. VonMises stress in Salomé or where ever, it is important to postprocess the EQUI_ELGA_SIGM field and not the EQUI_NOUE_SIGM field; this is because the equiv_noue_sigm field is created by interpolating the results from equi_elga_sigm which can lead to imprecise results, especially where stress concentrations are high.
To further muddy the waters, more precise results from Aster are obtained by using quads/hexahedrons over triangles/tetrahedrons.
Kees has some more valuable information at:
www.caelinux.org/wiki/index.php/Contrib:...nt_to_solid_elements
Regards,
Claus<br /><br />Post edited by: Claus, at: 2010/08/29 15:17
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
- Joe Dowsett
- Topic Author
- Offline
- Junior Member
Less
More
- Posts: 22
- Thank you received: 0
14 years 3 months ago #4666
by Joe Dowsett
Replied by Joe Dowsett on topic Re:Linear vs Quadratic mesh
Thanks, I'll make sure I remember that!
A better understanding of the whole process is on my list, rather than the suck-it-and-see approach I've used so far...
A better understanding of the whole process is on my list, rather than the suck-it-and-see approach I've used so far...
Moderators: catux
Time to create page: 0.182 seconds