I am simulating methane - air combustion on Ansys Fluent. I use DO radiation model. For the first 700 iterations, the solution continuously converges. But later, the radiation residual suddenly diverges and i got the message " Divergence detected in AMG for do-intensity".
My meshing is really good enaugh and solution parameters are correctly established.
Any advice on this problem is really appreciated.
I've seen this kind of error before and still can't always explain why it occurs. Just a few thoughts:
- Can you pinpoint the location in your geometry where the problem starts, e.g. by looking at a the solution right before the divergence occurs? Maybe your mesh is good in general terms but has an issue at that location?
- If you can pinpoint the location, try remeshing that region of your computational domain, even if you don't see any problems with your grid... I know from previous experience that this works quite often.
- Also, do you use any kind of symmetries in your case? They can sometimes cause problems if you don't adjust the settings in the DO model.
- which reaction model do you use? if you use some kind of eddy dissipation model/concept, did you choose an appropriate model for the absoprtion/emissivity coefficients in the materials sections?
Good luck!
Thank you very much for your response.
I use non-premixed combustion model. It is very interesting that the problem with my solution starts somewhere before the fuel and air meet each other. I did not mesh that region densely, because it is not a combustion zone.
There is no symmetry in the geometry and wsggm- Domain based is used for absorption coefficient.
I'd suggest remeshing the area where the problem first occurs... it has worked for me before.
Dear Mr. Leicher and Mr. Büyükakın,
I don't have enough experience about this topic but I wonder the effect of under-relaxation factor (related about DO) on solution divergence? At the beginning, can we reduce it a bit in order to provide convergence and then increase it to 1 again? [ I assume meshing is good.]
In the cases of "spontaneous divergence" that I've had so far, reducing the under-relaxation factor would only delay the onset of the divergence. I observed the same effect when I increased the number of steps between two radiation runs.
- Badges
- Science topic