AVL FIRE; Problem in modeling gas spray

doki · October 1, 2009, 10:14

I am Using AVL FIRE v8 to model the methane gas injection into cylinder. the Injection pressure varies from 70 to 200 bars. cylinder pressure varies also in the range of 20 bars.
due to the small diameter of injector nozzle (say 0.2mm), small mesh cells should be generated in the nozzle hole periphery; and because of high injection speeds (typically 400m/s), a numerically unstable case arises.
So I was wondering if anybody have faced such a problem, and how to solve it.
I mean is there any ways other than generation of such a fine mesh at the injector nozzle area?
I would be grateful to help me.

harishg · October 1, 2009, 16:00

What numerical method do you employ? How do you apply your boundary conditions?

doki · October 3, 2009, 05:47

I have used Different numerical schemes and they have shown the same results as above, but the main scheme is SIMPLE.
the Boundary conditions for the injector has been set as Mass flow inlet, currently a constant value.

harishg · October 3, 2009, 10:29

SIMPLE is not a scheme but a numerical algorithm for pressure velocity coupling. I meant the numerical method employed for the discretization of the convection term in the Navier-Stokes equations. When you have high gradients in the flow, use of central differencing schemes can lead to numerical instability.

doki · October 3, 2009, 11:41

well I just got your question. Yes, I have used the central differencing schemes as well.
the Numerical instability encountered is sth related to CFL number; i.e. the mesh size should be so that regarding other parameters, this number should be sth. about 0.8
But this size of mesh (and also time step) is somehow impossible to work with. the total required time for calculations grows substantially also.
any hints would be appreciated.

harishg · October 3, 2009, 15:09

People have used finer meshes for high speed flows and I do not think this may be a big issue. Here are a few things to try.

1. Use upwinded/MUSCL scheme instead of central difference for convection terms. This may result result in numerical dissipation but we need to find an optimum way for simulating.
2. Do you use pressure based or density based solver? For many pressure based solvers, the Courant number requirment is usually co<=0.5. So you would want to perform your computations with smaller courant numbers.
3. Do you use implicit (crank-nicholson, backward) or explicit time stepping? Try to use implicit if possible.
4. Check that the boundary conditions at all the domain boundaries are consistent with the physical problem.

good luck

doki · October 4, 2009, 11:22

first of all I should thank you for your response
I will try your suggestions in the case setup. as the velocities are so high, effects of compressibility are important and so I have to use density based methods.
regarding time step, I will use the implicit method as well. here is a problem of time step order, which should be increased with better quality coarse meshes; while gaining enough resolution for the problem; which is a challenging task!
anyway, thanks so much for your contribution

Jaivee · July 8, 2010, 10:01

hi,
Im a beginner and am working on validation of cryogenic injection into ic engine. this same case was done in ansys cfx and i have to do it in fire but am facing problems
i hav take multi material with species transport
hav activated constant mass flow in inlet
as helium is injected into air in fluid prop i hav chosen air
but i dont know how to select injected gas as helium

also i want to use gaillard model but i dont know how define d constants value..

plz help me out..

October 1, 2009, 10:14	AVL FIRE; Problem in modeling gas spray	#1
doki Member Join Date: Apr 2009 Posts: 41 Rep Power: 17	I am Using AVL FIRE v8 to model the methane gas injection into cylinder. the Injection pressure varies from 70 to 200 bars. cylinder pressure varies also in the range of 20 bars. due to the small diameter of injector nozzle (say 0.2mm), small mesh cells should be generated in the nozzle hole periphery; and because of high injection speeds (typically 400m/s), a numerically unstable case arises. So I was wondering if anybody have faced such a problem, and how to solve it. I mean is there any ways other than generation of such a fine mesh at the injector nozzle area? I would be grateful to help me.

October 3, 2009, 15:09		#6
harishg Senior Member N/A Join Date: Mar 2009 Posts: 189 Rep Power: 17	People have used finer meshes for high speed flows and I do not think this may be a big issue. Here are a few things to try. 1. Use upwinded/MUSCL scheme instead of central difference for convection terms. This may result result in numerical dissipation but we need to find an optimum way for simulating. 2. Do you use pressure based or density based solver? For many pressure based solvers, the Courant number requirment is usually co<=0.5. So you would want to perform your computations with smaller courant numbers. 3. Do you use implicit (crank-nicholson, backward) or explicit time stepping? Try to use implicit if possible. 4. Check that the boundary conditions at all the domain boundaries are consistent with the physical problem. good luck Meysam110 likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
AVL Fire transonic simulation	Ainil	Main CFD Forum	1	August 24, 2010 16:53
Sod shock tube problem with real gas effects	ganesh	Main CFD Forum	0	October 17, 2006 02:22
Dense gas dispersion modeling	Alberto	Siemens	1	October 3, 2006 17:03
Modeling Flow/Saturation/Absorption in Fibers	Gene Dougherty	Main CFD Forum	0	June 6, 2003 15:49
FIRE 7.2 AVL	Simone Lazzara	Main CFD Forum	0	July 11, 2002 06:30

October 1, 2009, 16:00		#2
harishg Senior Member N/A Join Date: Mar 2009 Posts: 189 Rep Power: 17	What numerical method do you employ? How do you apply your boundary conditions?

October 3, 2009, 05:47		#3
doki Member Join Date: Apr 2009 Posts: 41 Rep Power: 17	I have used Different numerical schemes and they have shown the same results as above, but the main scheme is SIMPLE. the Boundary conditions for the injector has been set as Mass flow inlet, currently a constant value.

October 3, 2009, 10:29		#4
harishg Senior Member N/A Join Date: Mar 2009 Posts: 189 Rep Power: 17	SIMPLE is not a scheme but a numerical algorithm for pressure velocity coupling. I meant the numerical method employed for the discretization of the convection term in the Navier-Stokes equations. When you have high gradients in the flow, use of central differencing schemes can lead to numerical instability.

October 3, 2009, 11:41		#5
doki Member Join Date: Apr 2009 Posts: 41 Rep Power: 17	well I just got your question. Yes, I have used the central differencing schemes as well. the Numerical instability encountered is sth related to CFL number; i.e. the mesh size should be so that regarding other parameters, this number should be sth. about 0.8 But this size of mesh (and also time step) is somehow impossible to work with. the total required time for calculations grows substantially also. any hints would be appreciated.

October 4, 2009, 11:22		#7
doki Member Join Date: Apr 2009 Posts: 41 Rep Power: 17	first of all I should thank you for your response I will try your suggestions in the case setup. as the velocities are so high, effects of compressibility are important and so I have to use density based methods. regarding time step, I will use the implicit method as well. here is a problem of time step order, which should be increased with better quality coarse meshes; while gaining enough resolution for the problem; which is a challenging task! anyway, thanks so much for your contribution

July 8, 2010, 10:01		#8
Jaivee New Member Jaivee T Joseph Join Date: Jun 2010 Location: hyderabad, mavelikara, munich Posts: 1 Rep Power: 0	hi, Im a beginner and am working on validation of cryogenic injection into ic engine. this same case was done in ansys cfx and i have to do it in fire but am facing problems i hav take multi material with species transport hav activated constant mass flow in inlet as helium is injected into air in fluid prop i hav chosen air but i dont know how to select injected gas as helium also i want to use gaillard model but i dont know how define d constants value.. plz help me out..