boundary condition for flame in open/closed box

[kirkcapitanF]

My problem is that I dont conserve mass in my domain and I get strange wiggles in velocity profile.
I use finite volume with AB2 and FFT solver for nabla²p.
I have a laminar flame starting at the centre of a box with periodic BC in y and z direction. I have to work on x direction BCs to achieve the case open or closed.
p=dynamic pressure;T=temp.;m(mx,my,mz)=mass flux.

staggered grid
-------> x
_____
|cell_1|
mx0--- ---mx1
| T,P |
--------


a)case Open box in x dir:
__ periodic __
dT/dn=0 dT/dn=0
dm/dn=0 |0|1| imax|imax+1| dm/dn=0
dp/dn=0 dp/dn=0
__ periodic __

d/dn means respectively: mx(-1)=mx(0), mx(imax+1)=mx(imax)
T(0)=T(1), T(imax+1)=T(imax), p(0)=p(1), p(imax+1)=p(imax)

a)case Closed box in x dir:
__ periodic __
dT/dn=0 dT/dn=0
m=0 |0|1| imax|imax+1| m=0
p=0 p=0
__ periodic __

If I plot x-velocity or z-velocity at the centerline (at kmax/2 and jmax/2) from 1 to imax it has wiggles on the left of the box. If I check integral of div(m) in the box it is very big. Any suggestions? Thanks in advance to everybody.

 

[btrueblood]

In your model, does the flame (combustion process) occur across a single cell? Try spreading it out a bit (use finer grid at the flame front). Remember that a combustion process is an amplification, and will amplify any incoming numerical "noise".

 

[btrueblood]

Also, what is your final outflow condition? Don't just use a constant mass flow, use a sonic boundary to help prevent downstream noise from propagating back.

 

[kirkcapitanF]

Thanks a lot!! The flame cover about 4 cells on a total of 400 in main direction. I used a finer grid as you suggested and the situation improved. Unfortunatelly after certain time I see spurious noise in my velocity field (i should simulate laminar premixed combustion but these instabilities produce some turbulence). For the outflow condition I tried to use outflow convective condition with a Ubulk fixed. I will try to use the sonic BC you suggested me, but I have to have a look for its meaning. Because I use a level set for the flame probably there could be an issue with the reinitialisation process, but it is just a guess of mine. I say this because the velocity instabiities appears at the beginning in the starting point of the flame and than they spead around. Thanks again.

 

[btrueblood]

Kirk,

If you are pretty sure the instabilities are coming from the flame cells, then ignore my 2nd post (at least for now).

Are you "injecting" fuel into the grid - and is the fuel flow or amount updating with each iteration, in order to try to maintain a constant fuel/air mixture ratio? Then, the amount of fuel is coupling to the incoming air from last time step, and causing the instability. Try holding the fuel injection as a constant mass flow, and adjusting it from run to run.

If you are assuming the fuel is premixed, and just begins combustion at the grid location you specified, then your best bet is to continue refining the grid, especially near the ignition point. You may also want to try seperating the combustion into a seperate grid or "block" (if your solver allows) so that you can adjust solver gain/damping parameters somewhat independently of the main flow grid.