Learning Forum

[menkch]

Hello there !

The first picture is my problem, the second picture is my mesh in general.

As you can see on the following pictures, I did a simulation of a jet flame in a constant wind with the FGM combustion model and LES. I had a CFL under 12 in most cells and I was aware of the very tiny cells in the burner which could be a problem. I don't know if that explains why I have a zone that is not solved as you can see on the first picture (very abrupt temperature change between blue and higher temperatures where the mesh changes.

Could you help me understand what happened ? Possible causes could be mesh quality but I expect Ansys would have given me a warning.

Does this unsolved part make the well solved part that follows obsolete and/or incorrect ? 

 

Thank you for your help,

Menkch

 

[Rob]

That looks like the mesh isn't conformal at that position. Is there a reason for decomposing the domain like that? 

[menkch]

PPS : Very sorry to write again, but yes, I have looked very closely and you are right it is not conformal on some nodes.

However, why would that cause the quenching in the vertical region up the nozzle ?

I would imagine that it creates some kind of discontinuity between the two non conformal meshes but here it simply does not burn vertically... Would you be able to tell me more about this ? Finally, this non-conformity is caused, I think, by the fact that I meshed without refinements on ICEM and then I refined via "refinement boxes" directly on Fluent because ICEM was not very practical for refinements...

How could I have done this better ? The refinement part I mean ? Are there other software that would have worked better to refine or is there a better way to refine an already existing mesh in Fluent with the certainty that it will remain conformal ?

[menkch]

I meshed with ICEM CFD the whole domain "at once". The domain is not "decomposed" per say, the different grid sizes are the different refinements (the smallest being around the nozzle). Now, with ICEM, the use of O-grids creates this weird cross shape, but I did not think this would a problem, especially since I used a very similar mesh (though slightly less refined) with RANS and FGM for a flame without wind and it burned well (I have attached the pictures of the RANS similar mesh and the good results).

 

[menkch]

PS : The black part just shows the location of the flame (stoeichioetric mixture fraction)

[Rob]

If the non conformal isn't defined you'll have a wall at that location, that may or may not cause something interesting to happen. 

ICEM CFD isn't used as much now, and we're seeing good results with the poly meshes from Fluent Meshing. That removes the need to do the blocking, and also tends to give a signifcantly higher quality mesh than automatic blocking. 

[menkch]

Thank you so much, Rob, I am happy to understand the problem !