Multiphase equilibrium of H2O solid, liquid and gas

[Nick Wogan]

Hi,

My ultimate goal is to compute multiphase equilibrium of many gases and solid materials with Cantera. To begin, I'm simply considering H2O gas, solid and liquid to understand how everything works.

Attached is a simple example, that computes equilibrium of H2O at different pressures and temperatures, to see what phase is preferred. Over some temperature and pressure ranges, the results make sense.

However, for some temperatures and pressures the results do not make sense. An example is when I compute equilibrium for 1 bar and 1000 K, the resulting mole mixing ratios are as follows

T = 1000 K, P = 1 bar
H2O_gas    0.0e+00
H2O_ice    1.0e+00
H2O_water  0.0e+00

All H2O is ice, which is of course wrong, given that T is above H2O's critical point.

Why does this happen, and how can I fix it?

Thanks for your help!

Nick

[Ray Speth]

Hi Nick,

The default multiphase equilibrium solver in Cantera can generate erroneous results when the the temperature is outside the bounds where the thermo polynomials for one of the phases are valid (or well behaved). This is a known issue (see https://github.com/Cantera/cantera/issues/270). For calculating equilibrium at a known temperature, I think the most robust solution would be to construct the Mixture object so it only includes phases that are in bounds for the corresponding polynomial.

I was going to suggest the secondary equilibrium solver, which can be used by specifying the argument solver='gibbs' to the mix.equilibrate method. This solver enforces the temperature bounds for each phase, and does not have this problem. However, I noticed that it finds the wrong equilibrium at T = 270, P = 1 bar — it should be ice, but the solver finds a gas instead.

Of course, you aren’t going to be able to get correct properties for water for your example condition of 500 K and 300 bar, which is near the critical point, by using ideal gas and condensed phase models, but that may not be relevant to the system you’re actually interested in.

Regards,
Ray

​