Simulation of Cylindrical Outwardly Flame Propagation by Ember

[サイトウトウ]

Dear Ember Users

I am trying to use Ember to simulate a cylindrical outlwardly propagating flame.

(The flame propagates away from axisymmetric axis without straining effect along the flame surface).

For the detailed configuration, please check the attached file.

I am following the example_cylindrical_outward.py, and setting the strain rate to 0.

(I believe by doing this the flame could propagate prependicularly to the unburnt gas, exactly like the outwardly propagating flame.)

The code runs, but the results seems incorrect as the burning velocity overestimated the laminar burning velocity. (Fig.1)

I have checked the simulation results at each time steps, and I found the flame temperature increased over the adiabatic flame temperature as flame propagating. (Fig.2)

The combustion seems intensified, and thus the burning velocity become larger (than the laminar burning velocity).

Since I have not found relative configurations, I wonder whether what I am doing is correct or not. 

Regards,

Cui

[Ray Speth]

Hi Cui,

This flame is moving quite rapidly through the domain (~20 m/s with respect to the burned gas). With the regridding parameters you have set, which result in regridding every 10 microseconds, the flame is often propagating into a region where the grid is underresolved. This results in an artificially high rate of diffusion and therefore flame speed. I found setting the regridding to occur every 2 microseconds (by setting regridStepInterval to 2) helped quite a bit. You also have to let the flame propagate out to radii that are quite a bit higher before you will start to see the flame speed come back down toward the value for the unstretched laminar flame -- to get the stretch rate down to ~100/s you need the flame radius to be about 20 cm. Running Ember like this isn't terribly efficient, but it does give the correct behavior.

I observe the persistent superadiabatic temperatures as well. It seems to be something that develops in the post-flame region. I don't think it's having much effect on the flame propagation itself, but it does suggest there's some issue there.

Regards,

Ray

[Tongtong CUI]

Dear Ray,

Thanks for your advise very much.

Follow your suggestion, I tried to use a fixed grid to let the post-flame region resolved as well.

Consequently, the temperaure converged to the adiabatic flame temperature successfully, and a natural flame propagation was retrieved! (Fig.1 and 2)

It seems that the adaptive grid refinement reduce the grid points at the stationary post-flame region, and this region becomes underresolved.

 

In this simulation, I fixed the initial grid domain (0.0 m to 0.1 m) with spatial interval (dx = 1.d0-5 m), and the time interval was dt = 1.0d-7 s.

regridStepInterval used a fairly large value (like 1000000) to let the adaptive grid refinement do not work. The detailed configurations are in the attached file.

The simulation ran about half an hour. (not bad for me)

I have also tried to simulate an extreme case (with a tiny time step (1.d0-9 s) and refined the grid every time step). 

In this case, the superadiabatic temperatures can be suppressed and became close to fig.1 and fig.2, but not as good as these results.

Regards,

Cui

2024年6月22日土曜日 3:14:07 UTC+9 Ray Speth: