Specification of air as the ONLY working fluid - massfrac

[SRkumar97]

Hi,
I am trying to run a supersonic boundary layer case over a flat plate, with standard air as the gas.
Now the issue I am facing is wrt the mass fractions and gas_IDs of N2 and O2. I want specify clearly that there is only and only air as the gas.
Now, I found a case which has got N2 and O2 with mass fractions as that of standard air - the Spray-Multijet testcase. Here, the prob_parm.H has the mass fractions stored in the variables Y_N2 and Y_O2 respectively. And the prob.cpp and prob.H files call for these two variables in massfrac[N2_ID] and massfrac[O2_ID] (e.g. - Spray-Multijet and Spray-Conv)

My doubt is, whether these two IDs are pre-defined in PeleC? In certain cases I'm seeing that an additional mass fraction is also specified with a value of absolute 0.0, (e.g. amrex::Real massfrac[NUM_SPECIES] = {0.0}; in Spray-TG) . But I don't understand this additional statement. I am unable to find any value for NUM_SPECIES in the prob_parm.H.

Would be glad if someone could explain all this and inform what is the correct way to specify air as the only working fluid.

I also came across this case involving air -

https://amrex-combustion.github.io/PeleC/validation/validation.html#counter-flow-diffusion-flame

But I am unable to find a case with a matching name in the RegTests folder.

Thanks for reading!

[dmontgomeryNREL]

PelePhysics offers several chemistry mechanisms that include O2. In the GNUmakefile for the Spray-Multijet case, you'll see the line Chemistry_Model := decane_3sp, which points to this specific mechanism in PelePhysics. In the associated mechanism.H file, you'll find three species defined: NC10H22, O2, and N2, each with a pre-defined index ID for the mass fraction. This file also contains the definition for NUM_SPECIES.

If you want to run a case with air, you can set Chemistry_Model := air, then initialize the mass fraction in PeleC using the pre-defined index ID's in the associated mechanism.H file. The air mechanism has two species, O2 and N2. If you only want O2, initialize massfrac[O2_ID] = 1.0 and massfrac[N2_ID] = 0.0, then use corresponding consistent boundary conditions.

[dmontgomeryNREL]

The EB-InflowBC case injects N2 into a domain initially filled with O2. The left gas is O2 and the injected right gas is N2. The lines of code following the ones you are questioning in the prob.cpp use the PelePhysics equations of state (specifically eos.PYT2RE) to get the density, R, and the energy, E, of the left and right gases , respectively.

eos.PYT2RE requires the argument Y to be an amrex::Real massfrac[NUM_SPECIES], which in this case is an array with 2 elements. Before the eos calls, the left and right mass fractions are initialized to ensure that all elements are set to zero, and then one of the elements is updated to 1.0. Initializing variables in C++ is considered best practice, as failing to do so can lead to undesirable or undefined quantities.

[SRkumar97]

Ok! So in a case wherein there is no usage of eos(), I can skip these two lines of initialization for N2 and O2 (amrex::Real massfrac_l[NUM_SPECIES] = {0.0}; and amrex::Real massfrac_r[NUM_SPECIES] = {0.0};)
And type directly the massfractions of N2 and O2, to be read from the prob_parm.H file, correct?
e.g.
massfrac[O2_ID] = prob_parm.massfrac_O2;
massfrac[N2_ID] = prob_parm.massfrac_N2;

[dmontgomeryNREL]

Yes, it is fine to manually initialize each species to a known value. Just be sure to initialize every element of the massfrac. If you select a mechanism with multiple species, but are only using a couple of species, initialize every element to zero first. I hope this helps.

[SRkumar97]

Got it, thanks!

[SRkumar97]

Hi, I was able to do my shock simulation, under inviscid assumption using Gamma Law, which is the ideal gas equation for standard air, with R=287 J/Kg-K. Since my case has no kind of reactions and is standard air, Gamma Law works fine. Now I want to run a supersonic boundary layer case. Here, I am happy and well set with using the Gamma Law and Sutherland's law for dynamic viscosity. I am trying to run standard boundary layer flows like in other CFD solvers. However, PeleC doesn't seem to allow this combination of Gamma Law with Sutherland's law. The exec compilation throws an error for this combination. How do I use Gamma Law with Sutherland's law, then?

Note - Here I'm only interested in Sutherland's law for dynamic viscosity. Which will reflect in the values of shear stress. Rest of the transport coefficients are unimportant; can be assumed constant.

[SRkumar97]

Thank you very much, I am able to correlate and understand whatever you have explained through the suite of testcases.