CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Native Meshers: snappyHexMesh and Others (http://www.cfd-online.com/Forums/openfoam-meshing-snappyhexmesh/)
-   -   simple geometry with big problems (http://www.cfd-online.com/Forums/openfoam-meshing-snappyhexmesh/108488-simple-geometry-big-problems.html)

Tobi October 24, 2012 20:07

simple geometry with big problems
 
4 Attachment(s)
Hi all,

I meshed many things with snappyHexMesh the last one year and noticed one thing. If you have a pipe in an angle (like 30 degree) to your background mesh, you always getting bad cells. This cells cousing bounding problems with your turbulence model.

What do I simulate
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
First of all:
I am meshing and simulating burning cambers with the flamelet model. But for
now I am just using the rhoSimpleFoam solver so get a first solution.

In the first simulations there aren t any geometry parts with an angle of about
30 degress or something like that (like a pipe). Everything works fine and
after doings some optimization I have parts of pipes in the goemetry
having angles like 30 - 70 degree to the background mesh

Now I meshed again and get bouding epsilon. (For sure its a message I know)
but its not getting away and keeps one value.


What did I find out
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
I had a look at the timesteps after a few iterations (1000) couse I am getting
bouding epsilon every 4 - 5 iterationstep. The problem zone is clear. The
pipes cousing those problems due to bad cells (see attached picture)


What did I do to correct the mesh
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
1. refine those regions (surface refinement) but that s no solution - bounding epsilon
2. refine the surface and build closed STL and refined the surface and the region of the closed STL - bouding epsilon
3. tried to create layer in the pipes without success...

3.1 layers are not generated over the whole pipe - bounding epsilon
3.2 switch off all criterias but there are still regions without layers

4. rotate my STL file that the pipes has no angle of 30 degree ... they are parallel now.

--> well that works... but now the other regions has an angle of 30 degree to the
background mesh and for those I am getting bouding epsilon.
BUT the epsilon values of the pipes are very nice now



Simplification
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
After I realized the problem I build a simple geometry depends on two pipes.
One pipe is parallel to the backgroundmesh so that the cells ar very nice, the
other one is rotated by 30 degree (see attached picture).

If I simulate that simple pipe, I get a bounding epsilon as well. So I reproduced
the error. The bouding epsilon occures every 3-4 timestep and has a fixed value
after several iterations (see the attached pictures; residuals).





What did I do
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
1. refine the regions - boundin epsilon
2. layer with collaped on a few cells - bouding epsilon



Question
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
What else can I do to perform a better mesh in the pipe with 30 degree.
I am out of ideas


Turbulence Model and more
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
I am using k-Epsilon and did some more changes:

switch schemes to upwind.
get better continuity error by setting the relTol to 0,0001
relaxationFactor set to 0.01 for k and epsilon,

U,p = 0.2

-> bouding epsilon



Any help would be very nice


Thanks for reading and now I go to bed... its too late :)
Tobi

mamiha October 26, 2012 12:50

Hi Tobi,

I think your background mesh is the origin of your problem. I guess you are building one block with blockMesh enclosing the whole geometry. Wouldn't it be possible for you to build your background mesh in such a way that it follows better your geometry? Looking at your example with two pipes connected by an angle of 30 degrees, I think it would be advisable to have two blocks, one for each pipe...

Tobi October 26, 2012 14:24

Quote:

Originally Posted by mamiha (Post 388723)
Hi Tobi,

I think your background mesh is the origin of your problem. I guess you are building one block with blockMesh enclosing the whole geometry. Wouldn't it be possible for you to build your background mesh in such a way that it follows better your geometry? Looking at your example with two pipes connected by an angle of 30 degrees, I think it would be advisable to have two blocks, one for each pipe...


Hi,

yes I thought about that but the problem is not that geometry :)
I have a very complex geometry and in that I think its not possible to do it. So I am trying to solve that for that simliyfied geometry

Is there a tool (gui) for blockMesh? Always using salome for that! Tobi

Tobi October 26, 2012 14:40

1 Attachment(s)
Hi,

its me again,

did you think that it `s possible to mesh a geometry like in the attachement with several blocks? For each pipe one which is rotated in the direction to to pipe ?

Hmmm I think thats very very complicated to do it step by step with blockMesh,...


Any ideas?
Couse your idea to use several blocks which are orientated correctly to the STL would solve the problem I think.

Tobi

Tobi October 26, 2012 14:55

1 Attachment(s)
Hmm ... okay ...

I can build several blocks with salome but I think its not possible to merge those two blocks to use it with sHM? Or I am wrong?...

There must be a solution, couse I think many people using sHM for complex geometrys and they have the same problems. Hmmm :(
Well maybe I can not mesh my combustion camber with that tool or without bouding epsilon...

Other helpful suggestions would be appreciated.

Tobi

mamiha October 26, 2012 15:40

Hi,

Ok, I see the problem :)
I do not know of any GUI for blockMesh, sorry...
I think you're right, in your case it would really be complicated to do it step by step with blockMesh. Not sure if it's worth a try...
I haven't worked with Salome yet so I'm not really aware of its capabilites, sorry. But if you create several blocks by their own and merge them later on, I would guess that you will probably end up with some polyhedral cells connecting the blocks. As far as I know, you need to have pure hexa background mesh for SnappyHexMesh.
Concerning your problem, I do not really have a satisfying solution. I am currently also struggling with similar problem...

wyldckat October 26, 2012 19:34

Greetings to all!

GUI for blockMesh? Here's one: http://openfoamwiki.net/index.php/SwiftBlock :D

Best regards,
Bruno


All times are GMT -4. The time now is 07:30.