CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Trying to achieve supersonic flow in a pipe (

jpo May 28, 2010 16:29

Trying to achieve supersonic flow in a pipe

I am trying to create supersonic flow in a pipe. I have mass-flow-inlet and pressure outlet at exit. Pressure outlet setting should be ignored by the solver and extrapolated from the interior if the flow is supersonic. I don't have a nozzle at inlet. My Supersonic/Initial pressure setting at inlet (the static inlet pressure) is low enough, and mass flux is high enough, so velocity is supersonic at initialization.

Still, the flow stays subsonic. What is wrong?

jpo May 30, 2010 11:50

According to Fluent manual, "the flow must be initialized to a supersonic flow or it will simply choke and remain subsonic"

I tried setting my inlet to mass-flow-inlet, pressure-inlet and pressure-far-field. Always did FMG initialization and plotted the velocity field. I could never initialize a supersonic flow, where speed is supposed to be highest at inlet and decreases toward outlet.

Then I tried patching the velocity to supersonic values. Once iterations start, it returns back to subsonic.

I am running out of ideas...

tez May 30, 2010 16:44

You can keep your exit pressure as much low as your solver allows. That value may not be physically possible, but as you stated, pressure outlet is ignored by solver when it is supersonic flow. You can keep it 1 Pa (abs) or negative gauge pressure at the outlet. If your pipe is uniform, that alone could drive the solution, on the other hand you can patch supersonic values to prevent chocking.
Hope that helps.

alikami May 31, 2010 03:09

as earlier suggested , try zero exit pressure.

jpo June 1, 2010 20:09

Thank you; I tried negative/low outlet pressure, but it didn't work.

"A real-life supersonic wind tunnel, for example, will not "start'' if the back pressure is simply lowered to its operating value; the flow will choke at the tunnel throat and will not transition to supersonic. The same holds true for a numerical simulation: the flow must be initialized to a supersonic flow or it will simply choke and remain subsonic." (Fluent manual, Section 26.15)

Negative/low pressure at outlet may produce supersonic flow initially; Yet after sufficient number of iterations the flow converts to subsonic again.

alikami June 2, 2010 01:54

try 2 things

1. give zero ambient to see flow behavior
2. increase mass flow rate at inlet incrementally , and check flow behavior


jpo June 2, 2010 10:57

Dear Ali,

many thanks for your suggestions...

Set ambient to zero, do FMG initialization, run. Increment inlet mass flux, run again.
This is what happens:
Initially, outlet flow is supersonic; as soon as run starts to converge, it drops down to subsonic; because I increase the mass flux gradually, density increases everywhere. This is choking.

Supersonic flow should be very different: highest velocity is at inlet, gradually decreasing towards outlet and possibly becoming sonic (M=1) there, if the pipe is long enough (Fanno flow).

alikami June 2, 2010 12:48

for fanno flow in ur case , input conditions are super as flow move along the constt area duct , normal shock will appear at certain length beyond that flow will be subsonic... right ! if your duct length is less than the distance to normal shock flow, M>=1 exist else not possible.check this link also this solvesRegardsAli

alikami June 2, 2010 12:50

to add , if you increase inlet pressure till the location of normal shock is further the duct length you will have flow with M>=1...
try increase inlet pressure

jpo June 2, 2010 14:50

The position of the normal shock should depend on inlet conditions and pipe length. For fixed inlet conditions there exist L* such that if pipe length L > L*, normal shock will occur around the pipe exit. Keeping everything the same, if L increases, the position of the shock will move towards inlet and eventually, supersonic conditions become impossible.

All this reasoning confirms your arguments. I read more about it here:

You are suggesting that increasing the inlet pressure should "move" the cutoff length L* further from inlet. Why is that? Can you elaborate more? Appreciate your thoughts.

alikami June 3, 2010 03:46
link above gives a good description

I hope it works for you.

jpo June 3, 2010 09:01

Ali, thank you very much for the link. It is a very nice and detailed treatment of supersonic flows in a pipe.

So far, I used "no slip" walls for my pipe in Fluent. Now if I set "zero shear stress" on all walls, it means the pipe length factor 4fL/D will become zero, because f = wall_friction_coefficient/rho*v^2

Then I should be able to see supersonic flow no matter how long the pipe is.
I will try this and see how it works.

kkostas1982 June 3, 2010 17:30

mixture in T
I would like to make a mixture of two identical liquid (water) with different densities from two different entrances, which will lead to a common output.
it could be done in laminar flow? would help too if you receive even a simple instruction. I'm new to the program and try to find a solution to this problem.

jpo June 3, 2010 17:56


yes, it works now that I have "perfectly smooth" walls. The link you posted is a GNU-licensed free textbook and is really helpful.

Now I can continue and work with "no slip" walls and examine things in detail

jpo June 3, 2010 17:59


I guess it depends what is your Reynolds number.

Please read

And also the thread
looking specifically at "laminar", "turbulent" and "Reynolds number" there.

All times are GMT -4. The time now is 19:48.