Supersonic Inlet for a Compressible Flow in a Nozzle
Hi!
I am trying to solve a compressible flow in a diverging converging problem. I am using the mesh given in the cornell problem: https://confluence.cornell.edu/display/SIMULATION/Fluent++Compressible+Flow+in+a+Nozzle The difference from the cornell tutorial is that I want to solve it for a Supersonic flow at the inlet. as if it was a diffuser. the area ratio A_inlet/A_t=3.5 I use Pressure Inlet with: P_total=101325, P_static=3733.8, T_total=765.2 which are the A/A* values for a the 1D isentropic problem. for the outlet I use Pressure Outlet with:P_static=3733.8, T_total=765.2 I am trying to reach supersonic inlet to sonic throat to supersonic outlet. I am solving it in steady state. the problem is that the program completely diverges after a few iterations and keeps on diverging. I tried refining the mesh with "Adapt" tool but it doesn't seem to help. What am I doing wrong? I have searched all over and I don't seem to find info about how to solve a problem with a Supersonic Inlet. maybe the problem is in the steady state? thanks! 
Which solver diverges first? That may be a hint. I've had divergence problems before and if I underrelax the solver that diverges first a little bit (~0.05 to ~0.1 change) it helps.

Thanks for the tip. I'll be sure to try it.
I'll post what happens when I get to it. 
Sure. It may take more a more aggressive relaxation effort than a 0.1 change so you can try that as well.

hi,
i want to put sinusoidal pressure inlet, is it possible in fluent? 
to your question wissou22, yes you can do it in fluent.
One way I know of is by UDF (User Defined function). check out Tutorial 4:Modeling Transient Compressible Flow in the FLUENT 12.0 tutorial guide. they do exactly what you want, only they do it at the outlet. NormalVector, I am solving the problem in a densitybased solver and I don't see an option for inputing relaxation parameters anywhere.. also at that subjet, If I'm solving a problem that should be supersonic (or sonic) in all the domain, What is the best solver to use? I used the density solver with its default settings: Formulation: Implicit Flux Type: RoeFDS Gradient: Least Squares Cell Based Flow: Second Order Upwind Are those settings good for supersonic flows? Thanks 
Hi, Daco.
I'm doing a similar work as you did. :) My simulation begins with 1st order Upwind scheme. It works well. But when I use 2nd order Upwind scheme, it diverged. Some people tell me, there isn't flux limiter in FLUENT. So it must diverge because of the numerical oscillation when using the 2nd upwind scheme. However, I don't how to get a 2nd order result in FLUENT. And they suggest CFX to me, since it has resolution scheme. Maybe the problem can be solved in CFX. The geometry I used in my work is similar to Anderson's. And his report is available as follows. http://ntrs.nasa.gov/archive/nasa/ca...1970018205.pdf http://ntrs.nasa.gov/archive/nasa/ca...1971017509.pdf If you get some improvement, please let me known: slaonslaon@gmail.com . Thanks. :) Quote:

Well, I have a partial answer because I left this project.
I found out that for the implicit solution method the program is completely divergent. When I tried Explicit it actually worked without exploding but it also didn't converge and the error monitor always moved around 10^1. So I suggest using the explicit method for supersonic inlets. Next, I think I had some problems with my geometry so I think some of my problems were caused by that. I reached a conclusion that because the geometry begins at the inlet without any straight geometry with a uniform flow before it, when a shock wave started to form at the inlet opening the program couldn't converge. so here I suggest when dealing with a condivnozzle use an initial straight geometry before the walls start to converge. Maybe even create a mesh outside the nozzle to see if a shock wave forms on the lips of the opening. Like I said I left this project and didn't reach a good converged solution. All I said here is only my conclusions while working so don't take it as a 100% truth. Isn't it silly that its very hard to solve a very simple case in a large commercial CFD program like FLUENT? I really don't get it.. Thanks! 
Thank you for your reply and suggestion.
:) 
All times are GMT 4. The time now is 00:35. 