|
|
nearest face centre. Additionally cells with face or point on the boundary get an explicit correction to pick up the exact nearest point on any point-connected boundary faces. |
|
|
|
I think all cells should be corrected, not just the wall cells. This is important for turbulence modeling. Otherwise, it is not going to be accurate enough. |
|
|
user313
2013-04-29 17:45
~0002138
Last edited: 2013-04-29 17:48
|
Instead of using "nearest face centers and then correction", why not use some other searching algorithm to find the true distances directly?
|
|
|
|
The current algorithm is a walking algorithm. The nearest wall point to any cell is the nearest point that any neighbouring cell has. It is very fast, handles transformations (rotational cyclics) and other coupled faces (processors). Any other algorithm (e.g. octree) will have (not unsurmountable) problems with these. |
|
|
|
I am testing to see how big the difference would be on the flow simulation. The wall distance difference in my simulation however shows that max(CurrentDist-TrueDist)/TrueDist=15%. This is not small. |
|
|
|
FYI. There is indeed difference in grid convergence study. Using the true distance gives closer results with other CFD codes. |
|
|
|
Interesting. Which turbulence model are you using? |
|
|
|
two, one is SA, the other is SST, the models I have modified to make sure the comparison is valid the models and b.c. are correct. |
|
|
|
For future reference, this is related to this report: http://www.openfoam.org/mantisbt/view.php?id=968 |
|
|
|
Resolved in OpenFOAM-dev, see
http://www.openfoam.org/mantisbt/view.php?id=968 |
|
