Setting T and P for PureFluid heptane fails at low temperatures

[speth]

Cantera version

2.4.0 and 2.5.0a2

Steps to reproduce

Run the following in Python:

p = ct.Heptane()
p.TP = 204.4, ct.one_atm

Generates the exception:

CanteraError thrown by Substance::set_TPp:
no convergence for P* = 6.19745e-07, V* = 0.257196

Setting the state to a temperature of 204.5 or higher works fine. The minimum temperature for the model is supposed to be 182.56 K.

[ischoegl]

I eventually ran into this issue while working on #907 (Substance::Tsat should raise an error for a pressure lower than the triple point pressure, which can be found by probing the triple point temperature); curiously, all other representations of tpx objects work down to the minimum temperature (I.e. the triple point temperature).

The Substance::set_TPp method could likely be replaced by a more conventional root finder (there appear to be quite a few parameters with ‘tuned’ switches that are difficult to interpret and may cause hung loops), but there’s also the possibility of an issue with property data (e.g. a typo) causing the failure documented in this issue.

[ischoegl]

Fwiw, I checked the values of numerical constants against the original source, and couldn't find a discrepancy (other than a minute difference for the value of Pc used in equation S-2, which appears to be inconsequential).

[ischoegl]

Making a note here that #907 deferred some fixes due to this issue (marked as @TODO):

• interfaces/cython/cantera/test/test_purefluid.py: in test_saturation_near_limits(self)
• src/tpx/Sub.cpp: in double Substance::Tsat(double p)