Low density Jet Instability

Assasin · February 23, 2016, 16:38

I need to model a helium jet being fired in ambient. Such flows are globally unstable and are seen to oscilate, one powerful validation is the frequency of oscilation.(published results are available.)
In the simulations till now i have been using a 3d grid with preconditioned real gas navier stokes solver in CFD++,inplicit 2nd order schemes for temporal,spatial discretization.I have tried with a lot of different grids but i have not managed to match the frequency except one grid. The animation for this 'correct' solution is attached, though if i further fine this mesh or corsen it (for mesh independence) , the frequency changes by a huge amount.
The reynolds number is 1800 and the video shows the helium concentration.
I am running unsteady solution after 1000 steps of steady state iterations.
kindly suggest what could be the problem and further course of action.

Thank you,
Anshul

FMDenaro · February 23, 2016, 17:43

The Re number is enough high to generate a broad range of scales of turbulence, this case can be afforded with DNS if you have enough computational resource (you should consider the smallest scales where reaction acts) or, alternatively, using LES with suitable model for the chemical reaction

Assasin · February 23, 2016, 23:46

Sir,
it is a laminar cold flow , no reactions are used.

FMDenaro · February 24, 2016, 04:36

Quote:

Originally Posted by Assasin

Sir,
it is a laminar cold flow , no reactions are used.

No, the flow is turbulent, the Re number based on the diameter is quite high and you have to consider that along the streamwise direction it increases further

Assasin · February 24, 2016, 06:54

I am sorry, meant to say that no turbulence model is being used.

FMDenaro · February 24, 2016, 07:29

ok, what about your grid resolution? If you were unable to use a suffiently fine grid to perform a DNS, you are working with the so-called LES no-model. The fact you are not using an SGS model can or not give you a goos solution depending on many factors. For example, if you are using disispative schemes, the artificial viscosity can mimic the SGS model (like in the ILES). the grid size is a further parameter influencing the solution.
I suggest to do some statistical analysis, compute the energy spectra to see if you have numerical issues (for exemple an energy pile-up).

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Implement simple barotropic relation - derive density field from pressure field	GJM1991	OpenFOAM Programming & Development	6	May 8, 2016 16:43
SIMPLE algorithm instability at low viscosities	kooroshg1	Main CFD Forum	7	July 10, 2014 04:00
Low temp. (~ 300 K) turb. jet & JANAF problems	bart_weisser	OpenFOAM Running, Solving & CFD	0	February 11, 2010 19:50
IMPINGING JET ........... HELP!!!!!!!!	Amir Omoumi	Main CFD Forum	10	August 30, 1999 23:11
how calculate the density in low mach number?	Juhee Lee	Main CFD Forum	1	July 31, 1999 16:26

February 23, 2016, 17:43		#2
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,882 Rep Power: 73	The Re number is enough high to generate a broad range of scales of turbulence, this case can be afforded with DNS if you have enough computational resource (you should consider the smallest scales where reaction acts) or, alternatively, using LES with suitable model for the chemical reaction

February 23, 2016, 23:46		#3
Assasin New Member Anshul Lakhani Join Date: Jun 2015 Posts: 6 Rep Power: 11	Sir, it is a laminar cold flow , no reactions are used.

February 24, 2016, 06:54		#5
Assasin New Member Anshul Lakhani Join Date: Jun 2015 Posts: 6 Rep Power: 11	I am sorry, meant to say that no turbulence model is being used.

February 24, 2016, 07:29		#6
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,882 Rep Power: 73	ok, what about your grid resolution? If you were unable to use a suffiently fine grid to perform a DNS, you are working with the so-called LES no-model. The fact you are not using an SGS model can or not give you a goos solution depending on many factors. For example, if you are using disispative schemes, the artificial viscosity can mimic the SGS model (like in the ILES). the grid size is a further parameter influencing the solution. I suggest to do some statistical analysis, compute the energy spectra to see if you have numerical issues (for exemple an energy pile-up).