[harman79]

Hello all,

over the past 10 days or so I have been trying to solve a 2d nozzle simulation in Fluent 12. I ve tried all viscous models, both with default settings and decreased under relaxation parameters but all my attempts either failed to reach < 10^-3 ish residuals or gave me some odd results.

There are three pressure inlets (left top, bottom and centre) and one pressure outlet.

I have attached a few pictures to help you help me.. As you can see the model is perfectly symmetrical (double checked) and the mesh is quite fine (or so I believe).. However, all the simulations that did converge suggested flows that were convected towards one side of the nozzle (please see pics).

Any suggestions on why this odd convection occurs???

Thank you very much for your time..

Harry

[gmwsy]

I am not sure what it is in your case, but sometimes the solution is physically asymmetrical even with symmetrical boundary conditions

[harman79]

Thanks for sharing your thoughts gmwsy..

The funny thing is that I ve also tried several different nozzle geometries with various Ae/Athroat values and I still get the same results...

[-mAx-]

There is a turbulent phenomenon (Coanda effect) where the stream flows on a side, instead of remaining axial (symetrical), but in your case the max residuals <1e-3 sounds to indicate a non-converged solution.
Also the Mach-Number distribution is quite weird (max velocity should be at the minimal section), but as you are in transient and compressible solver, I won't comment your result (because of 0-experience).
From my own experience (inconmpressible flow), I had sometimes convergence troubles with Pressure-inlet/Pressure-outlet combination.
Try to switch Pressure inlet with velocity inlet, and reiterate.

[AndyR]

Sonic line seems to be at throat so thats ok.
Flow will accelerate in expanding section but if nozzle is over expanded peak mach will be "inside" nozzle with seperation down stream. Given coarsness of mesh away from center line that separtion would not be well predicted, numeric issues might start it asymetrically, then grid / numeric issues could take over.

I do LOTs of converging diverging nozzles.

Mostly I use GASP, Fluent is last resort. Make sure you are using the coupled solver and started first order for a while before going second order.

Good Luck
- Andy R

[harman79]

Max and Andy,

thanks to both of you for your replies. The early separation "inside" the nozzle was purposely seeked for..

I have tried both the coupled (density based solver) and pressure based solvers with the same unsuccessful results. Have always started first order and then onto second order..

I assume that Andy's comment suggests a better meshing away from center line (and early inside the diverging part of the nozzle). I ll try that... I ll also try using a velocity inlet as Max suggested. As for using GASP...well, I must first learn how to use FLUENT and then persuade the boss to get a license for another package... Thats a toughie...

Just for the record, I have now managed to get a "symmetrical" result using an axisymmetric approach. However, I am a bit worried about the validity of such an approach as I will be trying to actually construct this thing and make it work for real...

I ll get back if I have any better luck

Cheers again for the help...

[Kruzster]

can sumbody provide me with a tutorial for -
1] modelling of flow thru an orificemeter ( i am especially interested in meshing..as i have a constraint of 10000 nodes for the total pipe)
2] how shuld i go abut solvinf in fluent..??

[gcd2006]

Originally Posted by harman79

Hello all,

over the past 10 days or so I have been trying to solve a 2d nozzle simulation in Fluent 12. I ve tried all viscous models, both with default settings and decreased under relaxation parameters but all my attempts either failed to reach < 10^-3 ish residuals or gave me some odd results.

There are three pressure inlets (left top, bottom and centre) and one pressure outlet.

I have attached a few pictures to help you help me.. As you can see the model is perfectly symmetrical (double checked) and the mesh is quite fine (or so I believe).. However, all the simulations that did converge suggested flows that were convected towards one side of the nozzle (please see pics).

Any suggestions on why this odd convection occurs???

Thank you very much for your time..

Harry

I also found the problem.
The nozzle supersonic condition can not be reached.

[gmwsy]

Originally Posted by harman79

Thanks for sharing your thoughts gmwsy..

The funny thing is that I ve also tried several different nozzle geometries with various Ae/Athroat values and I still get the same results...

Hi Harman,

What do you mean by same results?

According to my experience, it is usually not easy to get converged solution with both pressure inlet and pressure outlet.

Maybe you can try to use velocity inlet and see if the results will be improved. If it is the case, then the problem is the boundary condition rather than your mesh or model.

Good luck.

Shengyi

[flashkicker9]

Originally Posted by AndyR

Sonic line seems to be at throat so thats ok.
Flow will accelerate in expanding section but if nozzle is over expanded peak mach will be "inside" nozzle with seperation down stream. Given coarsness of mesh away from center line that separtion would not be well predicted, numeric issues might start it asymetrically, then grid / numeric issues could take over.

I do LOTs of converging diverging nozzles.

Mostly I use GASP, Fluent is last resort. Make sure you are using the coupled solver and started first order for a while before going second order.

Good Luck
- Andy R

hi, i am doing work on conveging nozzle with pressure farfield,could you tell what conditions to be taken for compressble flow to get the results in 3d as well as 2d models?

plz plz could you reply fast,
million thanks

[abdul Sami]

will u please tell me where u get nozzle dimensions any refrence plz???