[Zibi]

Unfortunately, if I increase the overall endTime, the cumulative error is increasing successively and sooner or later the simulation explodes. Additionally, I obtain totally wrong results (velocity and pressure are constantly increasing). As Alexeym stated, it's the fault of bad mesh quality (under-determined cells and non-orthogonal ones).

Hence, I'm stuck at my work right now... Cannot progress without a proper mesh.

Zibi

[ashishmagar600]

Hii..
seems you have been stuck on with meshing with stl files. I (and many of experienced foamers) recommend meshing with foam utilities for simulations with OpenFOAM. sHM is a very good tool for meshing, also cfMesh.

Anyways, stl files can be edited (ascii formats).

If you would like to learn using sHM there are numerous tutorials on the internet, even from OpenFOAM conferences and official foam developers. Just find out some, or else just pop in a text here, and I will post some links in the thread for you.

Regards.

[alexeym]

@Zibi

It is not that I am trying to convince you to use OpenFOAM, yet you have missed certain points in my previous post.

1. What if you miss something, while generating mesh? In tutorials they generate meshes with boundary layer, mixed hexahedral meshes (http://www.vmtk.org/tutorials/CenterlinesGmsh.html). What if you can produce higher quality meshes with VMTK?

2. You can generate mesh with snappyHexMesh initially producing mesh with single boundary defined by STL. Later you can use STLs (for inlet, outlet, etc) to create faces sets with topoSet utility and then use createPatch utility to create corresponding patches.

3. It would be great to have examples of your STLs, so people can try to propose more concrete solutions and not just "use snappyHexMesh".

[Zibi]

@ashishmagar600

That's what I was afraid of... The majority of 'easy-to-handle' tutorials I found in the internet (just a few of them thou) were basing on the several STL files (separate inlets/outlets/walls, etc), whereas I've got only 1 entire surface. Haven't found a tutorial how to explode the STL file into several patches (I just know that Salome MIGHT be able to do so). Hence, any material that could help me is warmly welcomed.

@alexeym

I'd really like to use the OpenFOAM software, since it offers a lot and I can see future in its development. Concerning points you have raised:

1. I tried several techniques of mesh generation: different constant edgelengths, different radius-adaptive edgelengths, adding boundary layers, etc. I tried exporting the meshes as Fluent ones (before and after tetrahedralization) and I tried meshing with Netgen and via the way you posted a link to. The latter method (with the use of gmsh), ended with gmsh freezing, despite numerous attempts. All of my tested meshes with the corresponding log files are located in the RAR file (see link below). Numerous attempts in getting finer mesh with VMTK resulted in its crashing/freezing...

https://www.dropbox.com/s/wxg4ufi5uv...eshes.rar?dl=0

2. I guess I'll end up with using the sHM, however, as stated above, I don't know how and where to start. I've got a single STL file that has to be exploded into separate STLs (inlets, outlets, walls). How about STL file that represents only wall surface? I mean a surface, where all the inlets and outlets are not present - is the open STL file somehow better/worse to operate on in sHM? In the file I posted a link to, there is a folder containing STL files.

3. As stated above - STL files are inside the posted link. The STL file called 'only_wall' wasn't used for any purposes - I just wanted to show you the opened STL file representing only the wall - can it be used in OpenFOAM or not-waterproof STLs are rubbish?

Zibi

[dickcruz]

Originally Posted by Zibi

@ashishmagar600

That's what I was afraid of... The majority of 'easy-to-handle' tutorials I found in the internet (just a few of them thou) were basing on the several STL files (separate inlets/outlets/walls, etc), whereas I've got only 1 entire surface. Haven't found a tutorial how to explode the STL file into several patches (I just know that Salome MIGHT be able to do so). Hence, any material that could help me is warmly welcomed.

...

2. I guess I'll end up with using the sHM, however, as stated above, I don't know how and where to start. I've got a single STL file that has to be exploded into separate STLs (inlets, outlets, walls). How about STL file that represents only wall surface? I mean a surface, where all the inlets and outlets are not present - is the open STL file somehow better/worse to operate on in sHM? In the file I posted a link to, there is a folder containing STL files.

3. As stated above - STL files are inside the posted link. The STL file called 'only_wall' wasn't used for any purposes - I just wanted to show you the opened STL file representing only the wall - can it be used in OpenFOAM or not-waterproof STLs are rubbish?

Zibi

Hello Zibi,

To address your STL issue, if you have access to the original model, you could use Salome or FreeCAD to split the STL. I'd say that FreeCAD has less of a learning curve.

For your snappy question, the tutorials are a good place to start. If you run a case, you can take a look at the mesh in paraView and analyze the system/snappyHexMeshDict file to see how it's built.

Remember to watch your units!

-JD

[ashishmagar600]

@Zibi

Can you share your stl geometry here?
Guess some of us can try helping.

Regards.
Ashish


edit:
Sorry it is already provided. my bad.

[danny66@supereva.it]

Hi,
I have an STL file representing a naca64 split in half. When I build the mesh with snappyHexMesh the box and the STL are perfecly merged, but when I execute simpleFoam and I launch the ParaView to make drag and lateral thrust forces, I got inconsistent results. The procedure I follow is this:
- Alphabetical extract blocks --> then I select my stl name --> apply
- Alphabetical extract surface --> apply
- Alphabetical generate surface normals --> then I click on Create Cell Normals --> apply
- Calculator --> cell normals --> Lateral thrust --> p*normalsZ --> apply
- Calculator --> cell normals --> Drag --> p*normalsX --> apply
- Alphabetical integrate variables --> apply
- then I select Cell Normals on the form


The result is:
the lateral thrust is 10 times more than drag ...but my naca64 split in half is symmetrical!! Why lateral thrust??
What I've noticed is: when I apply the computing of lateral thrust, my stl becomes orange in only one half. The other half is light cyan.

Anyone can tell me where I'm wrong?


Thank you all.