CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   Memory Problem (http://www.cfd-online.com/Forums/openfoam/95276-memory-problem.html)

ChrisPro December 12, 2011 08:33

Memory Problem
 
Hello!

I'm trying to mesh a rather complex STL-file. When i put this file in the middle of a "windtunnel" where the boundaries of the surrounding box are far away from the file, everything works fine and the caluclations seem to be realistic.

But in my case, it is necessary that at least one face (inlet patch) touches the stl, because I want to simulate the flow inside of an airfoil.
But when I make these adjustments I get the following error whilst running snappyHexMesh:
Code:

Edge intersection testing:
    Number of edges            : 8189103
    Number of edges to retest  : 8060091
new cannot satisfy memory request.
This does not necessarily mean you have run out of virtual memory.
It could be due to a stack violation caused by e.g. bad use of pointers or an out of date shared library
Aborted

So is there a way of calculating this by needing less virtual memory?
(I reset the swap size of my computer to ten times of the original size)

It doesn't matter how long the calucation takes. It simply should work!

thanks
Christian

wyldckat December 17, 2011 19:17

Greetings Christian,

I saw your post a while back, but only now have I been able to log in and try to answer you:
  • snappyHexMesh usually needs at least two things (besides the STL): a bounding box (usually generated by blockMesh) and a point that indicates the side of the mesh.
  • AFAIK, the bounding box must envelop the whole STL geometry, without touching it...
  • or you must supply feature edges that define the places where things get separated. Source for this statement: start reading from post #18 of thread http://www.cfd-online.com/Forums/ope...tml#post335295 and keep reading!
  • mmm... an airfoil...
    Quote:

    I want to simulate the flow inside of an airfoil.
    If I understand you correctly, then why not only mesh the inside of the geometry? You'll still need a bounding box nonetheless, in order for snappyHexMesh to be able to distinguish inside from outside.
Best regards,
Bruno

ChrisPro December 18, 2011 05:11

hey bruno!
thanks for your reply!!
I tried the same case on a computer with 24GB of memory and it worked :)

but thank you for your hint using featureEdge!!


And I also thought about just meshing the inside of course, but the problem is that the stl-file of the wing is hollow and the solid layer is very thin. So snappyHexMesh won't stop when reaching the boundary of the airfoil.

Or is there any option which handles this?

christian

sail December 18, 2011 11:38

Quote:

Originally Posted by ChrisPro (Post 336218)
hey bruno!
thanks for your reply!!
I tried the same case on a computer with 24GB of memory and it worked :)

but thank you for your hint using featureEdge!!


And I also thought about just meshing the inside of course, but the problem is that the stl-file of the wing is hollow and the solid layer is very thin. So snappyHexMesh won't stop when reaching the boundary of the airfoil.

Or is there any option which handles this?

christian

Have you tried to refine the mesh initially created by blockMesh in the area where the solid layer causes problems?

You can do that with the first step of refinement (ie, bounding box). Practically you need the cells in that area to be smaller of the solid thickness to let the snapping part to work and understand which cells have to be eliminated and to which side snap onto.

ChrisPro December 23, 2011 08:31

Ok thanks!
I'm going to try this!


All times are GMT -4. The time now is 21:41.