Reference Enthalpies
Enthalpy is a relative term. In order for different property packages to be compatible, they need to have the same reference enthalpy.
An example of a similar concept is temperature. Celcius and Kelvin have the same scale (if you heat water up by 1 degree Celcius, that's the same as saying you have heated water up by 1 degree Kelvin) but different reference points (1 degree Celcius is 276.15 degrees kelvin)
This problem also happens in property packages. If I have a Helmholtz property package an the enthalpy of water is 100 j/mol, it's at around 0 degrees celcius at atmospheric pressure. However, in a diferent property package an enthalpy of 100 j/mol might be 25 degrees celcius at atmospheric pressure. That's cos the reference point of IWAPS95 is the triple point of water (0.01 degrees C at 611 pa) while another one might use room temperature and pressure.
Our milk property package and our water property package have different reference enthalpies, as shown by this script (you can checkout at this commit and run it)
The script first prints the enthalpies at 300 kelvin and 1 atm, and then the enthalpies at 400 kelvin 1 atm.
python find_reference_enthalpy.py
milk enth_mol -281870.6061410932
helm enth_mol 2029.5082087775857
difference to add to milk: 283900.11434987077
--------------- vapor version ----------
milk enth_mol -238374.21957664166
helm enth_mol 49187.111188126015
difference to add to milk: 287561.33076476766
As you can see, there is a difference between them of ~283900. However, at the higher temperature, this difference increases to 287561. This is a difference of 3000 J/mol. Since it takes 75 J/mol to heat 1 mol of liquid water at 1atm 1 degree, that's kinda a lot.
However, if I modify the temperature to 350 kelvin, the difference in enthalpies decreases to 276482 J/mol. thats a "difference between the differences" of -7418, so it seems to vary quite a bit. Which is frustrating, because a reference enthalpy is a constant value you're supposed to add, and so that makes it hard to determine the constant to add.
Solution
To better understand what was going on, I tried to graph the result.
As you can see, the milk property package and the helmholtz/steam property package look quite different. Helmholtz has a sharp line at 100 degrees celcius, where everything vaporises. Milk doesn't have that at all- it slowly vaporises.
It turns out that this was because in the previous example we had the amount of water set to 0.999999999 - which is too high, and the formulation starts to break. Dropping off a couple of those nines turned out to give a much better result. Then, we just had to adjust the reference enthalpy of both the liquid and vapor in the milk_config file. (note that the reference enthalpy of vapor is actually the H parameter in NIST, and you have to put in a negative value for some reason - otherwise you'll end up with an inverted graph.)
The updated version can be found here.
Using this enables the Translator block to constrain enthalpy instead of temperature, which works a lot better.
Backlinks