I'm trying to solve a simple nozzle flow using CFX 10 and am having trouble with convergence!!. I'm sure its all due to the boundary conditions I'm using. Please could someone one give me some advice on the most robust boundary conditions for this problem. I've tried inlet velocity specified with an outlet static pressure specified, also inlet velocity specified and a supersonic outlet. The flow is definitely choked. The speeds are high so the steady state time step is very small 1e-07, hence a transient run would probably take weeks do solve. I've tried solving a subsonic case with specified inlet velocity and outlet static pressure and this solves fine for steady state. Any tips guys?

If the flow coming up to the throat is accelerating from subsonic to supersonic after the throat then Total pressure inlet/static pressure outlet should work fine.

You might have to play around with under-relaxation factors to attain convergence. Ask your support rep about how to relax gradients.

Cheers for the advice. I was playing last night with total inlet pressure and static outlet and having a bit of success. I think to get these residuals down I'll definitely need to adjust under-relaxation factors.

Hi CFX Guy,

If it is choked and you know the total conditions at the inlet, apply total pressure and total temperature at the inlet. At the outlet a static pressure lower than the critical pressure will give you choked flow, but you might have some funny behavior right at the outlet if the flow accelerates above Mach 1 through the throat. If that is the case, you can change the outlet to supersonic.

To "Name". There are no relaxation factors to meddle with in CFX, playing with the timestep is the most you would have to do. For this case, it should behave quite well, so setting your timestep to the residence time will suffice. A good check for the timestep is to plot streamlines in Post on a backup file and look at the average or maximum time on the streamlines.

Regards, Robin

In case of supersonic flow we always use

EXPERT PARAMETERS:

max continuity loops = 2 END

works every time!

Robin, generally I agree with you. With Ptotal inlet/Pstatic outlet you should only need to tweak the timestep a bit to get a converged answer.

However, there are under relaxation factors which can be tinkered with if needed to stablise things (not that this is always needed, only occasionally on very specific cases). For example, I've used these to help get through tough spots, especially on something like a converging/diverging nozzle with a shock, in the past:

ADVECTION SCHEME:
Option = High Resolution
Gradient Relaxation = 0.1
Blend Factor Relaxation = 0.1
END

Cheers guys i've managed to solve the nozzle case and capture the shocks. I didn't actually change the max cont loops to 2. Does changing this speed up to solution?

Am i right is saying that setting fixed inlet velocity for a choked flow case is not advisable? Since setting total pressure allows both the density, velocity and static pressure to change at the inlet as the solution develops.

Just try it and compare convergence plots (and CFPU time). Shouldn't be a very long simulation.