CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Convergence and Solver Selection (

cfduser9 January 9, 2012 23:38

Convergence and Solver Selection
I have created a gas distributor with high back pressure and a choked inlet distributing into a vacuum (the operating pressure is not quite zero, but very close).

Using the pressure-based solver, I have attempted to solve. The model always diverges. No matter how much I play with the under-relaxation factors, it simply won't converge.

Should I be using the density solver over the pressure solver? I am aware of the relative merits of each, but don't know what to do for this situation. The flow is slow near the inlet, but definitely compressible by the exit. It seems to fall between the two solvers, but maybe one is better than the other?

I should mention that the model is very complicated, as the distributor has many exits. A simpler model does converge using the pressure solver, but the real one does not.


duri January 10, 2012 03:38

Operating pressure doesn't matter for pressure based solver, if it is zero it will take a negative value. When compressibility effects are large (discharging into vacuum) you need to move to density based solver.

LuckyTran January 14, 2012 01:34

The density based solver is preferred over the pressure based solver, especially for compressible flows. The density based solver is more complicated and costly compared to the pressure based solver but is much more physically accurate.

The pressure based solver can still be used but requires a lot more care. If you are using the pressure based solver, did you remember to account for the viscous heating effects caused by the shear?

cfduser9 January 22, 2012 14:21

What do you mean by this? I did turn on the energy model and did then chose low-pressure boundary slip because it is exiting into what is essentially a vacuum. Is this what you are referring to?

LuckyTran January 22, 2012 14:34

You should be using the density solver.

If you want to use the pressure based solver, in addition to simply 'turning on energy' you need to account for viscous heating effects.

All times are GMT -4. The time now is 14:39.