Sonic jet
 

I'm analyzing a sonic jet. The results can't get converged. It always gives me a message "Absolute pressure limited to 1.000000e+00 in 37 cells on zone 2". I am using pressure inlet and pressure outlet. Any one have experience with this? Can you give me some suggestion? Thanks a lot! Joy

Re: Sonic jet
 

if the pressure ratio between the inlet and the ambient is too high (such that Mach disks are formed in the flow) the simulation might crash. it is advisable that you restart the simulation but increase the pressure ratio gradually. the better initial guess will help.

cheers nandu

Re: Sonic jet
 

Nandu, Thanks for all the help! Yeah, you are right, the pressure ratio between the inlet and the outlet is very high. But outlet pressure is high vacuum, not the ambient pressure. In the real situation, we can set the inlet pressure very high, and the outlet connects to a large reservoir where the pressure is pretty low. In FLUENT, what should we do in order to get the resonable result?

Re: Sonic jet
 

you can set your oulet pressure to be pretty much the same as what you would do experimentally. But i would set the inlet pressure such that the ratio of the pressures is pretty low (starting from 1). once you have perfectly expanded sonic jet, you can increase the inlet pressure ratio's in stages (typically factors of 2) until you reach the desired pressure ratio. Depending upon the number of shcok cells you wish to capture you can set your outlet boundary to be the the appropriate distance from the inlet. unlike other commercial codes the flow does not have to completely supersonic or subsonic at the outlet. Fluent can handle a mixed boundary. Use the Spallart-Allmaras turbulence model initially. I presume you are looking to achieve a steady state solution?

cheers nandu

Re: Sonic jet
 

Thanks Nandu,

I did what you said before. Somehow it helps. But in some cases it's still difficult to get the solution converged, beside sometimes it gives me the message "reversed flow in XX faces on pressure-outlet". What doest that suppose to mean? I also studied the effect of the dimension of the outlet chamber, of course, if should be wide and long engough. When I further extent the length, the flow field parameters remain the same. It looks like it's indenpendent of the length once it gets long enough. Do I expect to get different number of shcok cells for differnt lengths from outlet to nozzle? Currently, I am using the laminar model. Should I try Spallart-Allmaras turbulence model first and then switch to laminar model? what's the pros and cons of using Spallart-Allmaras turbulence model? Joy

Re: Sonic jet
 

1) you will have reversed flow through the outlet (pressure based outlet). this cannot be avoided, but it simply means that there is some mass flux entering throught the outlet. has no effect on the jet flow field

2) the width and length of the chamber is purely dependent on what you are trying to model. a sonic jet typically has a core region (where the flow is supersonic) bounded by a mixing region (the shear layer of the jet boundary) and a fully mixed region where the flow is then wholly subsonic. if you are only intersested in the first few shock cells, then limit the length of the domain to about +- 15 diameters. this will allow you to use the adaptive grid refinement and obtain high resolution simulations.

3) the first two (approximately) shock cells are for all practical purposes inviscid. the turbulence in the jet is bound within the jet boundary and will only manifest itself into the mean flow downstream of about 2-3 shock cells. i have used the spallart 1-eqn model, based on my literature survey and have found it to be quite accurate in determining the positions of the shocks in the flow. if the mixing of the jet is what you are interested in then you might want to try using a RANS based turbulence model.

hope it helps cheers nandu

Re: Sonic jet
 

Nandu, thanks very much for your help. I'v done some simulation using two differnt viscous Models and got the similar results. If you like, I can send you the results and let you help me to check the reliability of both methods. Thanks again, Joy.