Negative viscosity/conductivity from Chung corrections with SRK EOS

[baperry2]

The NonIdealChungCorrections routine for Simple transport provides modified transport coefficients for real gasses (i.e. SRK EOS in Pele). The corrections are based on empirical relationships from Chung 1984 and 1988 and sometimes can cause spuriously negative values of viscosity and thermal conductivity to be generated. Right now, we just clip any computed values for these quantities to 0.

Based on the sleuthing of @oneteaches2learn, we now have a better handle on the cause. Briefly, the correlations contain a term involving a mixture averaged reduce dipole that negatively influences the correction. This is estimated from the mixture-averaged critical temperature and volume, which are in turn estimated from the mixture-averaged Lennard-Jones parameters, computed from individual species Lennard-Jones parameters. For very polar species (only H2O, HF those relevant in our chemical mechanisms), the estimation of the reduced dipole in this manner is inaccurate and leads to a spuriously large value that can cause the computed viscosity and conductivity to be negative.

The proposed fix is to compute mixture-averaged critical temperature and volume directly from species critical parameters, rather than through the Lennard-Jones parameters. For species where we have real values of the critical properties, this would reduce inaccuracy in the Chung correction procedure. For species where we do not have this, we would estimate the quantities based on the Lennard-Jones parameters. The actual modifications will occur in NonIdealChungCorrections in Simple.H, but the quantities that get pre-computed when initializing TransParm will also need to change and we will need to modify CEPTR to allow for the critical volume to be output.

An alternative fix would be to use Lennard-Jones parameters computed from the critical properties of each species, rather than the actual parameters from the mechanism, only in the NonIDealChungCorrections computation.

This branch of PelePhysics contains a test that can be used to verify that a potential fix has removed the issue of negative transport coefficients: https://github.com/baperry2/PelePhysics/tree/chung-neg-visc-test