CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   Flame gets flushed away in reactingFoam (http://www.cfd-online.com/Forums/openfoam/93732-flame-gets-flushed-away-reactingfoam.html)

megacrout October 25, 2011 05:00

Flame gets flushed away in reactingFoam
 
Hi,

I am simulating a premixed flame with reactingFoam. I use methane with oxygen-rich air and a semi-developped chemistry (4 reactions, one reversible) with pretty unrealistic but still plausible kinetic data (however the problem is unlikely to be found here).

I set the inlet temperature to 800K but the mixture does not ignite. So I set the initial internal temperature to 2000K. The flame does ignite but the heat does not seem to diffuse fast enough so that the 800K mixture slowly replaces the hot gas and the flame disappears. Using a smaller inlet velocity does not solve the problem. I actually first tried with 1m/s and this should work.

I reduced the mesh down to 5x5mm cells. Courant numbers are perfectly fine.

Does it have sth to do with the solver? I looked pretty intensively at the equations and donīt see where the problem could lie (premixed gases should be fine).

Thanks for your help.

Tibo

dhondupant October 27, 2011 05:11

Quote:

Originally Posted by megacrout (Post 329312)
Hi,

I am simulating a premixed flame with reactingFoam. I use methane with oxygen-rich air and a semi-developped chemistry (4 reactions, one reversible) with pretty unrealistic but still plausible kinetic data (however the problem is unlikely to be found here).

I set the inlet temperature to 800K but the mixture does not ignite. So I set the initial internal temperature to 2000K. The flame does ignite but the heat does not seem to diffuse fast enough so that the 800K mixture slowly replaces the hot gas and the flame disappears. Using a smaller inlet velocity does not solve the problem. I actually first tried with 1m/s and this should work.

I reduced the mesh down to 5x5mm cells. Courant numbers are perfectly fine.

Does it have sth to do with the solver? I looked pretty intensively at the equations and donīt see where the problem could lie (premixed gases should be fine).

Thanks for your help.

Tibo

Have you tried 1D solvers with your 4 reactions chemical kinetic so far (like Cantera) or is your flame no laminar flame at all?

megacrout November 2, 2011 04:51

My flame is laminar but Iīm going to study turbulent flames as soon as it works with laminar ones. Plus I need to solve it 2D (I need both axial and radial profiles). So Iīd prefer sticking to reactingFoam.

Do you know how to solve this problem? or at least where this problem could come from?

Btw, I actually got good results with other kinetic data, so it seems that I was wrong assuming the problem could not be here. however, I do not know how to modify my data so that I get results out of it...

jenzkeller November 2, 2011 11:27

Quote:

Originally Posted by megacrout (Post 330406)
My flame is laminar but Iīm going to study turbulent flames as soon as it works with laminar ones. Plus I need to solve it 2D (I need both axial and radial profiles). So Iīd prefer sticking to reactingFoam.

Do you know how to solve this problem? or at least where this problem could come from?

Btw, I actually got good results with other kinetic data, so it seems that I was wrong assuming the problem could not be here. however, I do not know how to modify my data so that I get results out of it...

Well I had to model the laminar "Bennett"-flame using a self modified OpenFOAM solver.

In my case using any reacting mechanism with less number of reactions as in the GRI 1.2 was giving me bad results compared to experimental data.

So consider using much more chemical reactions for a laminar flame as for a turbulent flame.

You should also know that reactinFoam means physically Sc=1 using a unity Lewis number assumption which is absolutely inappropriate for laminar flames.

You might want to modify reactingFOAM too, by giving each species its own Lewis number.

cheers

megacrout November 2, 2011 12:33

Hi Jens,

Thanks for your answer too.

First of all, I am comparing OpenFOAM results with data from simulations obtained with other programms, not experimental data. So getting "real" data is not quite the problem even though non-conservation of inert species or non-sustainability of the flame obviously arenīt ok.
I managed simulating one-step reactions but it just does not seem to work with an intermediate number of reactions (for which I compute the kinetic data myself).

I read that detail about Sc=1 but I donīt understand where it comes from and - more important - what it implies. As for Le=1, I am not quite sure it still is the fact (from OF1.7.1 on) because of the commented line in hsEqn.H.

So, apart from the number of reactions and Sc numbers, do you know what I should have a look at to improve my results (i.e. get my flame to keep burning)?

Thank you very much.

jenzkeller November 3, 2011 04:24

Quote:

Originally Posted by megacrout (Post 330476)
So, apart from the number of reactions and Sc numbers, do you know what I should have a look at to improve my results (i.e. get my flame to keep burning)?

Well you can decrease the Courant-number which should not be above Co=0.2 or you can try to use a coarser mesh for an initial flame simulation.

cheers

megacrout November 7, 2011 06:03

Quote:

Originally Posted by megacrout (Post 330476)
Hi Jens,
I read that detail about Sc=1 but I donīt understand where it comes from and - more important - what it implies. As for Le=1, I am not quite sure it still is the fact (from OF1.7.1 on) because of the commented line in hsEqn.H.

I get it now: the Schmidt number is Sc = mu / rho / D. Setting it to unity means D = mu/rho which explains why the specieīs mass equation in reactingFoam (YEqn.H) contains mu instead of D to decribe the mass diffusion of the species!

However, I keep thinking a unity Lewis number does not apply to reactingFoam anymore. Can anyone help?

jenzkeller November 7, 2011 06:33

Quote:

Originally Posted by megacrout (Post 331019)
I get it now: the Schmidt number is Sc = mu / rho / D. Setting it to unity means D = mu/rho which explains why the specieīs mass equation in reactingFoam (YEqn.H) contains mu instead of D to decribe the mass diffusion of the species!

However, I keep thinking a unity Lewis number does not apply to reactingFoam anymore. Can anyone help?

Exactly.

I can advise you to read the first pages of Thierry Poinsot's "Theoretical and Numerical Combustion"

Than you will understand and see the simplifications done in the energy and species equation of OpenFOAM.

If you explicitly want to calculate Le=1 within your species equation just simply replace thermo->mu() with thermo->alpha().


All times are GMT -4. The time now is 04:37.