[Forrest]

I am trying to run a steady, implicit simulation using the SST k-w model with the geometry shown in the two pictures below. I am using symmetric boundary conditions to make use of the symmetry of the problem with pressure inlet/outlet boundary conditions. Currently I am using first-order discretizations of all the equations

I can't get the solution to go past the first ten iterations. I get warnings about the temperature, absolute pressure, and the turbulent viscosity being limited at different iterations. I also get warnings of reversed flow at the inlet and the outlet, but I have gotten by those before after the first 20 or so iterations in past simulations with different geometries. Final the code dies with the message:

divergence detected in AMG solver: temperature

I am unsure of what to change to get this going. I've tried running it with Laminar conditions, but it fails even quicker. I've also checked my mesh in GAMBIT to make sure all of the boundary conditions were applied correctly. Does anyone have any suggestions what I could do to get this to run? Thanks in advance

[-mAx-]

*do you get also problem if you disable energy equation?
*calculations with pressure inlet and pressure outlet are more difficult to get converged. An initial guess with the half of drop pressure may help.
*What is the max skewness of your mesh?

[Forrest]

*do you get also problem if you disable energy equation?

My solution does not diverge when I disable the energy equation. I've run it for around 100 iterations with out the energy equation disabled, and the only warnings that I got where that there was still some reversed flow at the inlet and exit. Although the number of faces where this occurred was going down as I iterated.

*What is the max skewness of your mesh?

GAMBIT shows that my maximum EQUISIZE SKEW is 0.954. I am thinking that my problems might be due to the size of my elements which have volumes that range from 2e-9 to 4e-16.

*calculations with pressure inlet and pressure outlet are more difficult to get converged. An initial guess with the half of drop pressure may help.

Should I change the boundary conditions for half of the pressure drop and run the simulation for a while and then increase my pressure drop?

[-mAx-]

ok, than it is linked to the energy equation.
Regarding the skewness, I don't think it is the reason, but I would recommand you to reach an mesh with skewness below 0.9.
For the pressure-BC, I meant, that you can initialize your solution with gauge pressure equal to the half of your drop pressure

[Forrest]

Ok thanks. I will work on my mesh and try the initialization idea you suggested.