natural gas VLE, thermodynamic model selection with Prode Software 1

[hcostello]

we have a realtime control on dew point specs based on Prode Software,
our gas includes C1-C6+ .. CO2, N2, H2S ... 0.005 H2O (no free water)
actually GERG 2008 is selected for VLE indicating water and some C6+ fractions as first components to condense,
comparing results with Peng Robinson differences are the range 1-4 K (and above)
while our max tolerance is 4 K ,
My understanding is that Peng Robinson is preferred by many companies but GERG 2008 should give more accurate values for densities and VLE (dew points) see the reports with chilled mirrors etc..
Prode includes many variants of Peng Robinson with specific alpha functions and different mixing rules,
according some calcs (comparing calculated dew points) these variants give values close to GERG 2008 (and different from Peng Robinson std.),
then my understanding is that these variants may be more accurate,
does this seem acceptable ?

 

[apetri]

I think that many US companies prefer PR while in Europe SRK has many followers
however simulators include variants of both PR and SRK and one can select,

for natural gas Prode Properties includes several models,
to name a few,

PRX which is the extended version of Peng Robinson available with van der Waals and complex mixing rules (Hiron Vidal, Michelsen, Wong-Sandler etc.)

standards as ISO 18453 or GERG 2008, both accurate for VLE

models including association term, as CPA (Cubic Plus Association) and SAFT

plus the usual models as SRK, LK, BWR etc..


selecting ISO 18453 or GERG 2008 you do not need VLE data regression (which may be required for models with complex mixing rules)

both ISO 18453 and GERG 2008 have specific ranges of application, outside these ranges you may adopt the other models listed above

as personal opinion, when possible I prefer CPA (CPA-PR-VDW or CPAE-PR-VDW selection) which allows to calculate hydrate phase equilibria.

Finally, to answer your last question, some models as PRX with complex mixing rules (for example PRX-NRTL-WSO) are very flexible and you'll be able to fit most VLE data sets,
but, as said, an advantage of ISO 18453 or GERG 2008 is that you don't need to go through the BIPs data regression procedure

 

[hcostello]

apetri,
thanks for the information,
I agree that GERG models are preferable due to "wired" parameters (while in other cases some VLE points and data regression may be required),
in my case I suppose that differences (PR vs GERG models) may be due to PR parameters,
I load the Kij from Prode database, do you suggest a specific data regression ?
However, if i select PRX-VDW model differences are much lower (with predefined Kij, from Prode database)

 

[apetri]

ISO 18453 includes specific parameters for PR alpha function,
PR standard in Prode may include up to 4 temperature dependent BIPs,
VLE data regression for BIPs may be useful,
however PRX model (the extended Peng Robinson in Prode Properties) has 4 (or may be more) parameters per each component which give accurate results for vapor pressure, density , enthalpy of pure fluids,
for mixtures van der Waals mixing rule may be good in most cases except with large amounts of polar fluids such has CO2, Water, H2S etc. etc.
for those cases with a EOS you may prefer a complex mixing rule,
I don't remember the limits of application for GERG 2008 but there are many documents showing that.

 

[hcostello]

thanks,
help appreciated