3 replies to this topic

[bponcet]

solving flash operations (on mixtures of hydrocarbons + water) I noticed large variations in predicted (liquid) water fractions, with different models (for example PRX vs. PR), 

for these models Prode database include only two BIPs (K12, L12) for many water-hydrocarbon pairs and no temperature dependence,

the data regression procedure allows to (re)calculate BIPs, do you think it would be advantageous in these cases ?

Should I select different models ?

[PaoloPemi]

for water-hydrocarbons phase equilibria the main differences (between Peng Robinson, PR and Peng Robinson Extended, PRX in Prode Properties) are due to the different alpha functions,

I think it is possible, within certain limits, reduce the differences including a temperature dependence in BIPs

but i do not see many advantages,

Peng Robinson (and Soave Redlich Kwong) are good for non-polar fluids,

for polar fluids there are alternatives (the Peng Robinson Extended in Prode Properties is an example)

[bponcet]

thanks Paolo,

I know that PRX has different alpha function, 

I have a client which does these calc.s with Kabadi-Danner (which is not available in my Base version),

these streams include hydrocarbons, N2, CO2, H2S, water (up to 8 % mole fract.), methanol etc.

which model do you suggest ?

[PaoloPemi]

Kabadi & Danner model (based on SRK) includes a composition dependence for water-hydrocarbons

aij = a1ij + a2ij * zw

a1ij = zi*zj*aij    a2ij = zi*zw*awij


in Prode the awij's  are icalculated only for water-hydrocarbons pairs (I believe the same in other simulators),

if you have other polar fluids (methanol etc.) better to use a extended EOS as PRX with complex mixing rules as Huron Vidal, in Prode Properties it is the PRX-NRTL (P-HV) model

in general I prefer PRX with complex mixing rules,
if BIPs are not present in Prode database, do use the data regression procedure...