[farzadmech]

Hello all
I am using snappyHexMesh for meshing my case and everthing seems fine except that I got 185 illegal faces(please see the figure attached illegalFaces.jpg), but when I do checkMesh it seems fine(no bad face), and I also can run my simualation with laminar flow(for Turbulent flow omega blows up since cells are really small).
Now, my question is that "Is my mesh good"? please see the checkMesh.jpg.

I found these two threads from simscale where they say that small amount of illegalFaces are ok;

https://www.simscale.com/forum/t/mes...faces/82276/17

https://www.simscale.com/forum/t/ill...reatment/68691

any help will be appreciated.

Thanks,
Farzad

[farzadmech]

One possible solution might be using castellated, snap and addLayers steps one by one as it says here;

https://develop.openfoam.com/Develop.../-/issues/2274

Again any help will be appreciated.


Thanks,
Farzad

Originally Posted by farzadmech

Hello all
I am using snappyHexMesh for meshing my case and everthing seems fine except that I got 185 illegal faces(please see the figure attached illegalFaces.jpg), but when I do checkMesh it seems fine(no bad face), and I also can run my simualation with laminar flow(for Turbulent flow omega blows up since cells are really small).
Now, my question is that "Is my mesh good"? please see the checkMesh.jpg.

I found these two threads from simscale where they say that small amount of illegalFaces are ok;

https://www.simscale.com/forum/t/mes...faces/82276/17

https://www.simscale.com/forum/t/ill...reatment/68691

any help will be appreciated.

Thanks,
Farzad

[Tobi]

Mesh can be okay. If you can run its fine. The omega problem might be related to something else. However, is you mesh scaled correctly? The volumes are almost very very tiny. If the volumes are fine, you should check, if the Knudsen Number is fine for continuum mechanics.

However,

• Running snappy in single steps might give another result as given by Mattijs in the issue you provided
• Furthermore, if one is deleting the base-length file, you will definitely get a different mesh

For me, even though, you did not provide the mesh in a visual way, its okay.

[farzadmech]

Dear Tobias
Thanks for your reply. I have attached my mesh and my solid inside the flow domain to this reply. I have used motorBike tutorial and I have tried to exactly follow that. My geometry is 2.4 cm and it has some attachment which has 20 micron diameter, but I do not understand why I should check knudsen number? Do you mean my geometry acts like a porous media? or what?

Thanks,
Farzad

Originally Posted by Tobi

Mesh can be okay. If you can run its fine. The omega problem might be related to something else. However, is you mesh scaled correctly? The volumes are almost very very tiny. If the volumes are fine, you should check, if the Knudsen Number is fine for continuum mechanics.

However,

• Running snappy in single steps might give another result as given by Mattijs in the issue you provided
• Furthermore, if one is deleting the base-length file, you will definitely get a different mesh

For me, even though, you did not provide the mesh in a visual way, its okay.

[farzadmech]

For the kOmegaSST problem(code blows up), I have calculated reference length scale based on length of the shape(2.4 cm) where I have very small diameters like 20 micron which is 1200 times smaller thank 2.4 cm, so I think Omega initial value cause the problem. I will tested it and let you know if kOmegaSST works or not.


https://www.openfoam.com/documentati...omega-sst.html

Thanks,
Farzad

Originally Posted by Tobi

Mesh can be okay. If you can run its fine. The omega problem might be related to something else. However, is you mesh scaled correctly? The volumes are almost very very tiny. If the volumes are fine, you should check, if the Knudsen Number is fine for continuum mechanics.

However,

• Running snappy in single steps might give another result as given by Mattijs in the issue you provided
• Furthermore, if one is deleting the base-length file, you will definitely get a different mesh

For me, even though, you did not provide the mesh in a visual way, its okay.

[farzadmech]

Dear all
Actually, I did a test and by increasing scale of the domain(1e6), I was able to mesh it without any problem and then I rescale the problem to the original shape using;

transformPoints -scale '(0.000001 0.000001 0.000001)'

Simulation wise, it seems wrong faces does not have meaningful effect on the results.


Thanks,
Farzad