[Q.E.D.]

Hello everybody!

this is my first post, so I hope I do everything right! :-)

A part of my thesis is to simulate the flow (incompressible, stationary) in a vacuum cleaner. Unfortunately I must keep the manufacturer and the geometry to myself because of the confidentiality clause.

The geometry is rather complex, so it took me a long time to create a "good" mesh. The geometry has been meshed with Snappy. My process of mesh creation was basically:

1. Import the STEP in Salome
2. Triangulation in Salome (NETGEN 1D 2D, Mefisto did not work)
3. Export of STL
4. Meshing with Snappy

I split the geometry into multiple pieces. (See Appendix) The single STLs are (except for the open surface, obviously) error-free. If I reassemble the single STL again, the surface is unfortunately still open. The geometry has very sharp edges in some places. Is it possible that some edges at this sharp tips connect to only one face?

After meshing with the appended SnappyHexMeshDict I get the appended CheckMesh Output.

As I said before, I have a steady and incompressible case. Therefore SimpleFoam ist used. Furthermore I use the kOmegaSST as turbulence model. The values ​​for k and Omega are estimated with Ferziger / Fluent. I use the University Cluster.

In the appendix you will find the residual plot, the log file of SimpleFoam (Slurm) and hopefully everything else that is needed. As you can see, the residuals plot shows a very strange behavior. Unfortunately, I can not explain it.

If the solver runs longer, the result doesn't change. I have already tried with LowReynolds wall functions. But then it diverges even faster.

What am I doing wrong?

Thanks for your support!!
Arthur

[mrurq]

Hello Arthur,

If you're having problems with the mesh I would suggest trying cfMesh. I have only used cfMesh a couple of times but friends have used it with poor stl files with good results. It's fast to install and for a first mesh you only need to specify the max cell size and input stl to get going.

If you need to install cfMesh go to http://sourceforge.net/projects/cfmesh/ and download the cfMesh-v1.1.1.tgz file. Save it anywhere you like.

tar -xvzf cfMesh-v1.1.1.taz
cd cfMesh-v1.1.1/
OF24x //Use the alias for the version you're using here!
source $WM_PROJECT_DIR/etc/bashrc .
./Allwmake

If your current mesh passes the checkMesh without problems then I would look at the first cells sizes and make sure they're size properly according to your turbulence model and wall treatment.

Regards,
Magnus

[Q.E.D.]

Hello Magnus,

thank you very much for your suggestions!

I've already downloaded cfmesh and found the tutorials quite useful. Nevertheless I failed to set up my case because the calculations brought no result. Maybe I should try again sometime...

Up to now I tried to resolve wall regions with at least 5 layers. Then I run simpleFoam to look at the Values of Yplus. Then I chose the right wall functions.

Thank you very much
Arthur