[adhiraj]

As per this bug report,
http://www.openfoam.org/mantisbt/view.php?id=327

OpenFOAM's thermophysical models have a problem when the thermodynamic database has different Tcommon for different species.
Has anyone tried to fix this issue?
For smaller mechanisms for hydrocarbons (C1-C2, ~ 30 species), I did not hit this issue, but for larger mechanisms including PAHs(~100 species), this seems to be pretty common.
Any help will be appreciated.
Thank you.

[dkxls]

Originally Posted by adhiraj

OpenFOAM's thermophysical models have a problem when the thermodynamic database has different Tcommon for different species.
Has anyone tried to fix this issue?

There is no "fix" to this problem in OpenFOAM.

The reason is related to the way OpenFOAM treat the mixture properties. The solution to the problem is to refit the polynomial data to a unified Tcommon. Here is a paper that might help you with the refitting (though their motivation for the refitting was different):
http://www.galcit.caltech.edu/EDL/pu...rmoDataNew.pdf

[dkxls]

Originally Posted by dkxls

There is no "fix" to this problem in OpenFOAM.

OK, actually there is a solution to that problem also in OpenFOAM. You would have to compute the Cp and h values explicitly for each species and average the species properties.
OpenFOAM uses an "implicit" averaging, I believe for computational efficiency reasons in the temperature calculation. Though, I haven't done any benchmarking.

I just remembered that I have implemented explicitly mass averaged transport properties (viscosity and heat conductivity), but I kept the implicit averaging for the thermo properties (Cp, h, ...).
I re-implement the "multiComponentMixture" class to accomplish that, it's rather straight forward for the transport properties but might turn out a bit more work for the thermo properties.

[adhiraj]

Thanks for replying.
I have also worked on implementing mixture-avergaed transport properties and for that I had to access the Cp and hs values for the individual species.
However, the computation of the thermodynamic properties is used in many places (calculation of reverse reactions, hs to T calculation, and so on) and so trying to override the implicit averaging in janafThermo would be highly non-trivial to do, if not impossible.
I will look into the paper and see if I can get around this problem by refitting the thermo properties of the species I need.

[dkxls]

Originally Posted by adhiraj

override the implicit averaging in janafThermo would be highly non-trivial to do, if not impossible.

The implicit averaging is only partly implemented in the janaf class, actually the relevant part for you is in the "multiComponentMixture" class.
Thus, to implement an explicit averaging you just have to re-implement the "multiComponentMixture" class (i.e. make a new class).

Besides, do you really need the thermo properties of the mixture for revers reactions? I would have guessed that only species properties are used there.

To my knowledge, the mixture properties are only calculated in the thermo classes. Outside the thermo class the mixture data is only accessible via the respective fields or functions, something like this (in the top-level solver):

volScalarField& MuMixture = thermo.mu();

tmp<volScalarField> CpMixture = thermo.Cp();