[Tobi]

Dear Community,

I am happy and glad to tell you that I finished to analyze all my cases in the investigation of the "Influence of mesh size, orientation and type on numerical schemes in a convective flow field".

Information

• Investigate into numerical schemes to convective flow field

• Influence of schemes on mesh size

• Influence of schemes on mesh type (structured / unstructured)

• Influence of schemes on mesh/flow orientation (45degree / 0 degree)

• Cases analyzed: 108

• Available on my homepage / English / German

• Provide all *eps files for your research (if you need it)

• Provide the cases and a bash script for running all 108 simulations

• Provide a description page

• You find the stuff on menue OpenFOAM | Numerical Schemes - www.holzmann-cfd.de
  

I hope someone find this work a bit interesting and helpful.
If you want to send me some feedback do it.

Kind regards,
Tobi

[dkxls]

Hi Tobi,

nice study! (I just had a quick look at the figures though.)

At a first glace I don't see any surprises there, but it nicely illustrates the differences between first and second order schemes.

Best regards,
Armin

[Tobi]

Hi Armin,

thanks for the feedback. I know that there is nothing special but I thought for some beginners its very interesting. For guys like you, who are familiar to cfd, I think its not very helpful or show some new things but as you mentioned... just to have a look at some schemes and the dependencies to mesh parameteres, I think its nice to know.

[Tushar@cfd]

Good Work

[vatavuk]

Hi Tobias,
I was expecting the structured mesh to give better results than the unstructured one because of its regularity, but in your results generally both meshes have similar results and sometimes the unstructured is better.
What is the number of cells for the unstructured meshes?
I suggest that you also test a polyhedral mesh created from your unstructured mesh (using the polyDualMesh utility).
Best Regards,
Paulo

[markusrehm]

Hi Tobias,

have you also tested the impact of the "bounded" divSchemes?

e.g.

div(phi,U) bounded Gauss linearUpwindV grad(U);

Regards, Markus.

[Tobi]

Dear Markus,

no I did not check the bounded scheme because it was implemented later on. In previous FOAM versions this mathematical trick was implemented with -> Sp(phi, x). You can find some PDF I made for that here in some thread, but I don 't remember.

Maybe check out the libOpenSMOKE thread (here I suggest you to get to the FOAM wiki because I made a summary of that big thread).

[squick]

Hi Tobi,

Very nice job on the convective term.

I did something a bit similar but more for the Laplacian term :

http://www.paralgo.fr/publications/C...ce-OF-2015.pdf

I did start with a Stokes problem like in R. Pierre. In the link the results are shown for Naviers Stokes but in both cases problems of convergence appear on unstructured grids. This difficulty appears also on the wall driven cavity test:

http://www.cfd-online.com/Forums/ope...ed-meshes.html

I would be very interrested to know your opinion.

[Tobi]

Hi all,

I want to inform you that during the weekend I changed the section »Numerical Schemes« on my webpage as follows:

• Added a new mesh type (polygon) that was been investigated
• Added two new numerical schemes (OSPRE, QuadraticFit)
• Added the automatized study
• Updated the plots that can be downloaded (*eps and *png)
• Comparison between structured (0° and 45), tetraeder and polygon mesh is done
• Some typos and formatting

Have fun.