Can initial condition makes the solution diverge?

 Register Blogs Members List Search Today's Posts Mark Forums Read

July 28, 2019, 15:02
Can initial condition makes the solution diverge?
#1
New Member

Leegunho
Join Date: Jul 2019
Posts: 11
Rep Power: 4
Hello Foamer! I'm beginner of CFD.

I'm trying to simulate Supersonic nozzle with various Mach number.

The inlet pressure is 10bar, and outlet pressure is vacuum.

Inside of the nozzle is initially vacuum too.

Imagine the air is dumped into the vacuum tank from the pressure tank through the supersonic nozzle.

*The problem is that if i initialize the values to vacuum condition(P=0 or 1), solution diverges.

When i set the initial condition same as the inlet values(P=10bar), solution works.

But i want the processes of reaching steadystate from vacuum condtition, i want to set the initial values as vacuum.

I used Openfoam and Fluent, both of them diverged in vacuum initial condition. (SonicFoam, rhocentralFoam in openfoam)

If you have any ideas, please let me know.

Attached Images
 Boundary_Initial.JPG (46.3 KB, 55 views)

 July 30, 2019, 12:46 #2 Senior Member   Peter Baskovich Join Date: Jul 2014 Posts: 127 Rep Power: 9 Yes, initial conditions can and often do cause divergence (usually in combination with other factors). I would not set outlet to 0 as this is non-physical (bad start if you want real physics out). Which pressure are you setting? Can you share your solver output, it usually tells you what went wrong.

 July 31, 2019, 03:56 #3 Senior Member   Carlo_P Join Date: May 2019 Location: Italy Posts: 176 Rep Power: 5 Hey Kim, maybe the problem con be the relaxation factors. In a case so difficult, you can have maybe some numerical issues. At which iteration would diverge? Can you try with very small relaxation factor? Are you initializing all the domain with p=0? This can be a problem, since not all the domain in the result is p=0. If you want, you can try to initialize in different way, using TopoSetDict . Cheers, Carlo p.s. can you post your BC files? I'm really interested in it.