Moist air at very low temperatures

[thorade]

The nonlinear solver sometimes uses very low or even negative temperatures in function calls to h_pTX, resulting in errors, because of the log(T) in the h_Tlow function. I suggest to catch negative temperatures and return h=-inf for T<=0, will send a PR.

The following model was used to investigate and it revealed another problem:

model MoistAirTLowTest
  replaceable package Medium = Modelica.Media.Air.MoistAir;
  Modelica.Blocks.Sources.Ramp Tramp(
    height=-320,
    duration=0.8,
    offset=323.15,
    startTime=0.1)
    annotation (Placement(transformation(extent={{-80,38},{-60,60}})));

  Medium.AbsolutePressure pamb= 101325;
  Medium.Temperature T;
  Medium.SpecificEnthalpy h;

equation 
  T = Tramp.y;
  h = Medium.h_pTX(pamb,T,Medium.reference_X);
  annotation (uses(Modelica(version="3.2.2")));
end MoistAirTLowTest;

After simulation, plot h with T as independent variable. The bump in the curve might push the solver even further in the wrong direction, I believe.

[dietmarw]

Next time, no need to create the issue. Just send the PR :-)

[beutlich]

Next time, no need to create the issue. Just send the PR :-)

Did I miss something?

annotation (uses(Modelica(version="3.2.2")));

And not even up-to-date. :wink:

[dietmarw]

The PR is basically an issue in itself. If somebody like @thorade is so nice to offer a PR they can simply save the effort of creating a separate issue.

[beutlich]

will send a PR.

Ah, I missed it.

[thorade]

The PR will only address the safeguard for log(T) but not the little bump. Ah, and forgive the crappy commit message, only thought about it after pushing.

Also, not sure yet why my model calls it in that region, and why there are no errors from saturationPressure or sublimationPressure.

[beutlich]

Ah, and forgive the crappy commit message, only thought about it after pushing

You still can git commit -ammend and git push --force, but otherwise, if we squash merge ...