[fcollonv]

Dear foamers,

I'm trying to do the simulation used by Weller et al. to validate XiFoam
[H.G. Weller et al., 1998, Application of a Flame-Wrinkling LES Combustion Model to a Turbulent Mixing Layer, Twenty-Seventh Symposium (International) on Combustion/The Combustion Institute].

For that first I took simply the tutorial case, refine a bit the mesh at the shear layer and boundary layers and update the U and k file to use the experimental profiles.
Then the flame was effectively wrinkled (see XiFoamInit.jpg VS WellerResults.jpg - on the last the top picture is from the experiment). But the temperature and the pressure were crazy. In XiFoamInitTemperature.jpg I use a threshold filter to select the cells with a temperature between 260K and 290K. That range is totally unphysical as the temperature should not go below 293K.
So I changed the numerical schemes (cf. attachment). And now the pressure and the temperature are great: p is roughly constant and the minimal temperature is 292.6 K However the instability of the flame is gone.

Any suggestions will be appreciate.
Thanks,

Frederic

Additional information about the simulation:
* max CFL = 0.5
* no gravitation
* the mesh is not has large in the homogeneous direction as in the Weller's paper. The dimensions are those of the tutorial case
* the combustion properties are those of the tutorial case
* it looks like the wrong smaller temperature was triggering the instabilities due to a bigger density difference between the fresh gas and the burnt one.

[hannes]

Hello Frederic,

I have noticed the same behaviour of the Weller model as you when I simulated this testcase with OF-1.5dev some years ago and I did not find the reason for that.
I have implemented two other combustion models and tested them on the same testcase (same grid and BC's like tutorial case) and they yielded much more physical results.
Take a look here:
http://www.openfoamworkshop.org/08/p...nesKroeger.pdf

Also, I did not succeed in applying the weller model to a partially premixed bunsen flame case. The newton solver for temperature did not converge quite often and also unphysical temperatures occurred.

Regards, Hannes

[alberto]

Since the author of the paper is also one of the developers of OpenFOAM, a report of this issue on mantis ( http://www.openfoam.com/mantisbt/main_page.php ) might give you the answer and eventually a solution.

Best,

[fcollonv]

Thanks Hannes and Alberto for the quick answer.

@Hannes: When using the tabulated chemistry, do you read the density from the table? If so, the pressure used in the ideal gas law is the one set in Cantera (presumably 1 atm.) to compute the flamelet, isn't it? You were consequently using a kind of incompressible ideal gas law (as named in Fluent) in which one makes use of the so-called operating pressure independent of the local pressure. Can you confirm that?
For another combustion model, I implemented such incompressible ideal gas law to get rid of numerical acoustic effects as OpenFOAM has no perfectly non-reflecting BC. The improvement was important...

@Alberto: Thank you for the suggestion. I actually thought of it. But I was unsure as it isn't really a bug in OpenFOAM... I will give it a try.

[alberto]

Originally Posted by fcollonv

@Alberto: Thank you for the suggestion. I actually thought of it. But I was unsure as it isn't really a bug in OpenFOAM... I will give it a try.

I don't want to encourage users to report setup problems on bugzilla, but if you can't reproduce a result in the literature using the same model, it might be a bug. In the end, it's better for everyone to have this clarified ;-)

[fcollonv]

In the end, it's better for everyone to have this clarified ;-)

I agree totally

[fcollonv]

Hey guys,

Here is the answer from Henry Weller for the reported bug:

I have now spent some time running and re-running this case with numerous combinations of settings, meshes etc. and for much longer than I could afford to do in the mid-90s when I first ran the case and it is indeed problematic. The flame surface instability is VERY sensitive to the setup of the case, the boundary conditions, the numerical schemes, mesh etc. to the point that it is very difficult to use it as a reliable validation case. We will consider alternative cases both for validation and to provide a more reliable tutorial case for XiFoam.

[henrik]

Hi,

Inspired by the results of Tekgül (2017) who extended the inlet section, we revisited this problem and we added a ramp, which conforms to the original paper. The ramp is designed using potential theory and ensures development of the boundary layers before the step, which leads to the flame instabilities seen in the experiment. These cases can be found here:
https://develop.openfoam.com/committees/hpc/-/tree/develop/combustion/XiFoam/pitzDaily3D

The effort has been done as part of the exaFOAM project

https://exafoam.eu/
https://www.linkedin.com/in/exafoam-project/

Best Regards,
Henrik & Sergey

Tekgül, B. (2017). On the applicability of progress variable approach for large eddy simulation of premixed flames (Master's thesis, Middle East Technical University).