[DongMei]

Hi guys, I encounter one problem while running a case with CONVERGE. Following are the details.
The problem occurs when I run a case of propane-air combustion, the condition including mass flow rate, equivalence ratio and temperature, is same as that in our experiment in which no flashback happens. I test two mass flow rates, but flareback always happens shown in following picture.
mesh.jpg
The combustion model is SAGE.

The turbulence model is RNG k-e.

The ignition property is given in following picture.

When the mesh is refined, the predicted velocity over the nozzle exit (D=10mm) reduces, that's another question for me.
Thanks very much.

[DongMei]

Ignition property is shown as follows.
QQ??20180306162841.jpg
SAGE property is shown as follows.
QQ??20180306162828.jpg

[Gaurav_Kumar]

DongMei,

I have a few questions on your case setup.
Is there a centrebody in your geometry? Is the nozzle flow swirling?
I see that you are using a source to start combustion but source start time is 10s. Is that how long you are running your simulation or your start time is large?

At the conditions your ran, I believe, you were expecting a stable flame which would very close to flashback. After this you were looking to reduce the flow rate of air or increase fuel flow or add another fuel (like H2) to initiate flashback. If not, then could you help explain the setup.

What mechanism you are using?

Please take a look at the message I sent you at cfd-online inbox. I would be glad to take a look at your setup.

Originally Posted by DongMei

Ignition property is shown as follows.
Attachment 61883
SAGE property is shown as follows.
Attachment 61882

[DongMei]

Gaurav Kumar,
Very glad to see your reply.As for your questions, following are the details.
1. The nozzle flow is swirling. But CONVERGE suggests to use mass flow rate instead of velocity, so I wonder how the velocity of the flow is computed at the nozzle exit, especially swirling flow.The geometry is only a part of the test rig to save time. And I have attached the geometry. The boundary conditions of two sides are periodic.
QQ??20180307151034.png
2. Yes, the condition in the simulation doesn't witness flachbask in respective experiment. Now I am trying a case with increasing mass flow rate. It seems as the fule flow increases, flashback betters a litter but still happens.
3. I tried two mechanisms, one is two-steps shown as follows, the other is UCSD(sandiego20161214_mechCK.dat,http://web.eng.ucsd.edu/mae/groups/c...mechanism.html). It seems no difference appears.
Two-steps mechanism:
ELEMENTS
N H O C
END
SPECIES
N2 O2 CO CO2 H2O
C3H8
END
REACTIONS
C3H8+ 3.5O2=>3CO+4H2O 5.8232e+12 0.0 34000.0d0
FORD /C3H8 0.8/
FORD /O2 0.86/
CO+ 0.5O2<=>CO2 2.0e+09 0.0 12000.0d0
END

4. The reason for 10s is that the geometry needs to be filled with premixed propane-air, it takes some time. I set 10s before the iginition to have the space filled with premixed gas. I have no ideas if it affects the premixed combustion process and leads to flashback.
5. Sorry, I didn't see the message, please send it again.
6. At last, I'd like to repeat one question mentioned earlier: Why the predicted fuel velocity at the nozzle exit reduces when the mesh is refined? The flame speed seems higher than the gas velocity at the exit.

Best regards.

[DongMei]

Originally Posted by DongMei

5. Sorry, I didn't see the message, please send it again.

I'm sorry to say that because I receive your message in my email instead of the cfd-online inbox.

[khaledhmz]

Hello, did you find a solution to this problem, I have the same in my combustor...
need your help...