Gas phase mixture enthalpy values decreasing with increased pressure about thermo

[CalebBell]

Hi Odhran!

Yes, this is expected behavior. By the way, there is now a high-accuracy gas-only pseudo-pure model specifically for air in Thermo.

from thermo import *
import numpy as np
from thermo.phases import DryAirLemmon
import matplotlib.pyplot as plt
from thermo.chemical_package import lemmon2000_constants, lemmon2000_correlations
T, P = 298.15, 101325.0
gas = DryAirLemmon(T=T, P=P)
flasher = FlashPureVLS(constants=lemmon2000_constants, correlations=lemmon2000_correlations,
                   gas=gas, liquids=[], solids=[])
Ps = np.linspace(500000,20000000, 100)
gas_H = np.zeros_like(Ps)

k=0
for P in Ps:
    gas_H[k] = flasher.flash(T=500, P=float(P)).H()
    k=k+1

plt.figure()
plt.plot(Ps,gas_H, label = "gas_H")
plt.xlabel("Pressure")
plt.ylabel("Enthalpy")
plt.legend()
plt.show()

It also shows the same behavior. Note that when compressing air in real life, the temperature increases as the process is isentropic, which results in the significant compression duty.

[CalebBell]

Hi Odhran,
Unfortunately the Lemmon model is pseudo-pure, i.e. incapable of participating in vapor-liquid equilibrium. In thermo, you can only use it for obtaining vapor properties. Adding a liquid phase would not work with it. Additionally, it is slower due to its complexity.
Sincerely,
Caleb

[OdhranOOC]

Much appreciated Caleb! Ah interesting I did read that Lemmon paper a few years back actually. So can I swap out the current gas phase EOS then for the improved Lemmon one, and leave the liquid phase as it is? How would that look?

My current code for reference:

constants, properties = ChemicalConstantsPackage.from_IDs(['nitrogen', 'oxygen', 'argon'])
kijs = IPDB.get_ip_asymmetric_matrix('ChemSep PR', constants.CASs, 'kij')
eos_kwargs = {'Pcs': constants.Pcs, 'Tcs': constants.Tcs, 'omegas': constants.omegas, 'kijs': kijs}
gas = CEOSGas(PRMIXTranslatedConsistent, eos_kwargs=eos_kwargs, HeatCapacityGases=properties.HeatCapacityGases)
liquid = CEOSLiquid(PRMIXTranslatedConsistent, eos_kwargs=eos_kwargs, HeatCapacityGases=properties.HeatCapacityGases)
flasher = FlashVL(constants, properties, liquid=liquid, gas=gas)
zs = normalize([78.08, 20.95, .93])

Also, do you know if the Lemmon model is faster than the alternative I have above, or is it actually slower due to the increased accuracy?

Much appreciated