Problem with supersonic flow

BastianS · March 10, 2012, 07:32

Hello together,

i´ve got a problem with the Fluent solver that don`t converg. I´am simulating a radial turbine blade and want to show the pressure field with the shock waves.

I tried the following:

- different mesh to exclude that the error is caused by the mesh
- reduced courant number (0.1 ; 0.01 etc.)
- solution steering with supersonic flow in the "run calculation" settings
- velocity inlet instead of pressure far field in the boundary conditions
- w-cycle, v-cycle, F-cycle in the "advanced solution controls"

I always used the density based solver. Inviscid and energy equation is enabled.

Does anyone have an idea of how I can make the solver converge?

Kind regards

BastianS · March 11, 2012, 12:51

Does no one have an Idea?

robboflea · March 12, 2012, 12:48

What are the boundary conditions that you specify? How is your domain shaped?

Also, but this is just my curiosity, why have you chosen euler solver?

Rob

BastianS · March 12, 2012, 12:59

I used the euler equation after I have used the k-epsilon. I just wanted to simplifie the problem as a test because the solver didn't converge with turbulence model.

I mesh the geometrie with triangle or submap in Gambit.

Boundaries are:

INLET (pressure far field)
- Mach 1.9
- 140000 Pa absolute

OUTLET (pressure outlet)
- 20000 Pa absolut

After 3 days of try and fail I got greate improvements with the hybrid initialization (first I always tried standard initialization)

robboflea · March 12, 2012, 14:02

Why do you initialize the inlet as a pressure far field? I would try with pressure inlet.
Instead of using the hybrid initialization try to initialize the problem using fmg initialization (in the Fluent console you can find it under 'solve/initialize/FMG initialization').

Hope it helps,

Rob

BastianS · March 12, 2012, 14:31

Dear Rob,

thanks a lot for your reply. I´ll try out your suggestions.

Concerning pressure far field:

I found a tutorial in wich they used the pressure far field. So that´s why I used it too

robboflea · March 12, 2012, 14:39

Pretty curious, but it depends also on how is geometrically designed your domain.
Let me know if it works, normally FMG initialization solved many of my divergence problems.

March 10, 2012, 07:32	Problem with supersonic flow	#1
BastianS New Member Bastian Schammann Join Date: Mar 2012 Posts: 5 Rep Power: 14	Hello together, i´ve got a problem with the Fluent solver that don`t converg. I´am simulating a radial turbine blade and want to show the pressure field with the shock waves. I tried the following: - different mesh to exclude that the error is caused by the mesh - reduced courant number (0.1 ; 0.01 etc.) - solution steering with supersonic flow in the "run calculation" settings - velocity inlet instead of pressure far field in the boundary conditions - w-cycle, v-cycle, F-cycle in the "advanced solution controls" I always used the density based solver. Inviscid and energy equation is enabled. Does anyone have an idea of how I can make the solver converge? Kind regards

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Numerical Solution of the Blunt Body in Supersonic Flow Problem	andres17	Main CFD Forum	19	April 26, 2016 16:29
mesh generation for supersonic flow followed by freestream	padmanathan	FLUENT	6	September 14, 2010 08:53
supersonic flow problem	yekaniyasari	OpenFOAM Running, Solving & CFD	0	April 14, 2008 08:02
Multiphase flow problem	icedou	FLUENT	6	July 10, 2005 03:52
fluid flow fundas	ram	Main CFD Forum	5	June 17, 2000 22:31

March 11, 2012, 12:51		#2
BastianS New Member Bastian Schammann Join Date: Mar 2012 Posts: 5 Rep Power: 14	Does no one have an Idea?

March 12, 2012, 12:48		#3
robboflea Senior Member Join Date: Nov 2011 Posts: 109 Rep Power: 15	What are the boundary conditions that you specify? How is your domain shaped? Also, but this is just my curiosity, why have you chosen euler solver? Rob

March 12, 2012, 12:59		#4
BastianS New Member Bastian Schammann Join Date: Mar 2012 Posts: 5 Rep Power: 14	I used the euler equation after I have used the k-epsilon. I just wanted to simplifie the problem as a test because the solver didn't converge with turbulence model. I mesh the geometrie with triangle or submap in Gambit. Boundaries are: INLET (pressure far field) - Mach 1.9 - 140000 Pa absolute OUTLET (pressure outlet) - 20000 Pa absolut After 3 days of try and fail I got greate improvements with the hybrid initialization (first I always tried standard initialization)

March 12, 2012, 14:02		#5
robboflea Senior Member Join Date: Nov 2011 Posts: 109 Rep Power: 15	Why do you initialize the inlet as a pressure far field? I would try with pressure inlet. Instead of using the hybrid initialization try to initialize the problem using fmg initialization (in the Fluent console you can find it under 'solve/initialize/FMG initialization'). Hope it helps, Rob

March 12, 2012, 14:31		#6
BastianS New Member Bastian Schammann Join Date: Mar 2012 Posts: 5 Rep Power: 14	Dear Rob, thanks a lot for your reply. I´ll try out your suggestions. Concerning pressure far field: I found a tutorial in wich they used the pressure far field. So that´s why I used it too

March 12, 2012, 14:39		#7
robboflea Senior Member Join Date: Nov 2011 Posts: 109 Rep Power: 15	Pretty curious, but it depends also on how is geometrically designed your domain. Let me know if it works, normally FMG initialization solved many of my divergence problems.