CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Effect of Grids on Solution (http://www.cfd-online.com/Forums/main/13443-effect-grids-solution.html)

Andrea May 11, 2007 04:37

Effect of Grids on Solution
 
Hi,

If a CFD solution shows the footprint of the mesh (especially at the shared patches/edges in multi-block structured grids), what would be the reason and how can I fix it? This hapens only in some part of the domain, not everywhere.

Thanks, Andrea

to May 11, 2007 05:06

Re: Effect of Grids on Solution
 
Real CFD problem or just visualisation flaw ?

Mani May 11, 2007 08:16

Re: Effect of Grids on Solution
 
>shared patches/edges

Are these matching interfaces (e.g. each cell on block 1 connects to a cell on block 2), are they non-matching, overlapping (chimera).... ?

How are the interface conditions defined. Do you use ghost cells?

Here is what you can do as a check: Run a simple case (e.g. a 2D airfoil) on a single block. Then decompose the single block into a number of smaller blocks and run the code in multiblock mode. These two converged solutions must be identical!

Sue May 11, 2007 16:13

Re: Effect of Grids on Solution
 
Well, I assume it seems to be a CFD problem, as I use TECPLOT for visualisation, so that front should be ok. Right?

Sue May 11, 2007 16:14

Re: Effect of Grids on Solution
 
I agree with Sue.

Andrea

Andrea May 11, 2007 16:19

Re: Effect of Grids on Solution
 
This is not the first case that I run this code for, in previous cases it was great, the properties transfer very nice and smoothly between the blocks. Even in this case, it works well in other parts of the domain (other blocks). I am just wondering why this is being happened in some (but important) regions.

Andrea

ag May 13, 2007 21:34

Re: Effect of Grids on Solution
 
Do you have strong gradients in this region that don't appear elsewhere, or in your previous solutions? If so, you may still have problems in how you are transferring information that only show up when strong gradients are present. Some transfer schemes work well in regions of weak gradients and not so well in regions of strong gradients.


All times are GMT -4. The time now is 02:34.