Vapor Fraction / Quality should return NaN outside of vapor dome

[ischoegl]

Problem description

The quality of a saturated mixture is only defined under the vapor dome below the critical point (as the ratio of vapor mass to total mass); the definition could be used for compressed liquid or superheated vapor, but breaks down for supercritical mixtures.

Steps to reproduce / Behavior

In [1]: import cantera as ct
   ...: w = ct.Water()

In [2]: w.TQ = 300, 0
   ...: w.TV = 300, w.v*.9 # compressed liquid

In [3]: w.Q # should be NaN
Out[3]: 0.0

In [4]: w.TQ = 300, 1
   ...: w.TV = 300, w.v * 1.1 # superheated vapor

In [5]: w.Q # should be NaN
Out[5]: 1.0

In [6]: w.TP = w.critical_temperature + 1, w.critical_pressure # supercritical

In [7]: w.Q # should be NaN
Out[7]: 1.0

C's SNAN or C++'s std::numeric_limits<double>::quiet_NaN() could be used instead in the C++ layer, e.g.

double Substance::x()
{
    if (T >= Tcrit()) {
        return SNAN; //(1.0/Rho < Vcrit() ? 0.0 : 1.0);
    } else {
        update_sat();
        if (Rho <= Rhv) {
  	    return (Rho == Rhv ? 1.0 : SNAN); //1.0;
        } else if (Rho >= Rhf) {
            return (Rho == Rhf ? 0.0 : SNAN); //0.0;
        } else {
            double vv = 1.0/Rhv;
            double vl = 1.0/Rhf;
            return (1.0/Rho - vl)/(vv - vl);
        }
    }
}

The change will make it necessary to update logic for exception handling in some places; correct behavior also needs to be checked for MATLAB as well as various file IO formats for SolutionArrays - i.e. CSV/HDF (and pandas DataFrames).

System information

• Cantera version: all
• OS: any
• Python/MATLAB/other software versions: any

[ischoegl]

As a point of reference, CoolProp returns -1 outside of the vapor dome.

[ischoegl]

For all practical purposes, this is pretty inconsequential. If Cantera/enhancements#35 is tackled, things should be updated so handling is consistent.