I try to study the turbulence inside a vertical cylinder. In a first time, I take the case of steady flow. The air is introduced axially from below, deflected by means of a small conical deflector and thanks to the geometry of the chamber the air goes up into the cylinder. I take the ideal gas law, k-epsilon model, inlet and outlet pressure for boundary conditions.

My problem is that I have directly the message "divergence detected in AMG solver : k" when I want to iterate.

I try a lot of solutions I have find in this forum without any result.

Could anyone help me?

Thanks in advance

Check your hardware, especially RAM sockets. I have experience, that this could be initiated by some bad memory address sectors.

I had a similar problem, but with pressure correction. I found that I had some highly skewed cells. Once I corrected the highly skewed cells by adapting the iso-values of cell skewness my case began to iterate. Perhaps this may be your problem. First initialize the solution and then go to contours grid and select equiangle skew and click compute to get you min and maxs. Having skewed cells of 0.9 or higher isn't good, as in my case. Hope this helps

I have check the skewed cells but it appears that in my case, this number is lower than 0.4 Is it to high?

[SEDL]

Specify reasonable values of K and Epsilon in boundary conditions

[vinaynitrkl]

what about 1.3057*e^(-10) ??
is this too small?
i too got same error..

[SEDL]

Go thru this to calculate turbulence values:
http://support.esi-cfd.com/esi-users/turb_parameters/

[yoque]

u can find the problem when u look at the residuals of your solution. try to find crossections between residuals , then change under-relaxation factors. I think you can solve your problem like this.

[Mortie]

I had this problem,
in some cases I think the problem is because of Courant number and under-relaxation factors and even in some cases it is because of skewness in your mesh,
but in my case it was because of my mistake in boundary definitions, I mean I forgot to select one of the name selection edges when I was meshing, so take care and review your process of meshing before reducing CFL and URFs

[Arjun_kalkur]

I experienced the problem too.. but in my case it was because of few negative volumes, once you get rid of these.. I think you can iterate peacefully