3.

2 Thermodynamic Models
The Peng-Robinson equation of state [9], the UMR-PRU model [32, 39], and the PC-SAFT EoS
[40, 41] are evaluated. These models are well documented in the literature, so they are only
briefly discussed here.

3.2.1 Peng-Robinson EoS

The Peng-Robinson EoS [9] is described by the following equations:
RT a
P= − (3.1)
v − b (v − b)(v + b)
a = ac a(T) (3.2)

(RTc )2
ac = 0.45724 (3.3)
Pc
a(T) = [1 + m(1 − Tr0.5 )]2 (3.4)

m = 0.37464 + 1.54226𝜔 − 0.26992𝜔2 (3.5)

RTc
b = 0.07780 (3.6)
Pc
Tc and Pc are the compound critical temperature and pressure, 𝜔 is the acentric factor and Tr
is the reduced temperature.
For the extension of the PR to mixtures, the classic van der Waals one-fluid mixing and com-
bining rules are commonly used:
∑∑
a= xi xj aij (3.7)
∑
b= xi bi (3.8)
1
aij = (ai aj ) ∕2 (1 − kij ) (3.9)
where x is the mole fraction, kij in Eq. 3.9 is a binary adjustable interaction parameter of com-
ponents i and j, that is determined by fitting binary phase equilibrium data.
For the compounds investigated in this chapter, the critical properties (Tc and Pc ) and the
acentric factors (𝜔) are taken from DIPPR [42].

3.2.2 PC-SAFT EoS

One of the most successful attempts to correct the weaknesses of cubic equations of state is the
statistical associating fluid theory (SAFT) family of equations of state. SAFT is based on the
perturbation theory of Wertheim [43, 44] and was originally developed by Chapman et al. [45].
In this study, the Perturbed-Chain SAFT (PC-SAFT) proposed by Gross and Sadowski
[40, 41] is considered, which is one of the most successful modifications of the SAFT theory.
The main difference between original SAFT and PC-SAFT is the reference fluid used. Specifi-
cally, PC-SAFT uses the hard-chain reference fluid to account for the dispersion interactions
unlike SAFT, which uses the hard-sphere reference fluid. In the PC-SAFT equation, the residual
Helmholtz energy ares (T, V, N) is given as the summation of three terms, the hard-chain (ahc ),
the dispersion (adisp ), and the association (aassoc ), each one representing contributions from the
corresponding kind of intermolecular forces:
ares (T, V, N) = a(T, V, N) − aideal (T, V, N) = ahc + adisp + aassoc (3.10)
The hard-chain term (ahc ) is given by:
ahc ahs ∑
=m − xi (mi − 1)lngiihs (𝜎ii ) (3.11)
RT RT i

where mi is the number of segments in a chain of component i, m is the mean segment number
∑
in the mixture (m = i xi mi ).
The hard-sphere contribution (ahs ) is given by:
[ ( 3 ) ]
ahs 1 3𝜁1 𝜁2 𝜁2 3 𝜁2
= + + − 𝜁0 ln(1 − 𝜁3 ) (3.12)
RT 𝜁0 1 − 𝜁3 𝜁3 (1 − 𝜁3 )2 𝜁3 2
𝜋 ∑
𝜁n = 𝜌 x m dn (3.13)
6 i i i i
[ ( 𝜀 )]
di = 𝜎i 1 − 0.12exp −3 i (3.14)
kT
where 𝜎i is the segment diameter, and 𝜀i is the depth of the potential.
The radial distribution function of the hard sphere fluid (giihs ) is calculated as follows:
( ) ( )
1 di dj 3𝜁2 di dj 2 2𝜁2 2
giihs = + + (3.15)
1 − 𝜁3 di + dj (1 − 𝜁3 )2 di + dj (1 − 𝜁3 )3
The dispersion term is modeled using a second order perturbation theory on chain molecules
rather than hard spheres, using the following expression:
adisp
= −2𝜋𝜌I1 (n, m)m2 𝜀𝜎 3 − 𝜋𝜌mC1 (n, m)I2 (n, m)m2 𝜀2 𝜎 3 (3.16)
RT
where the integrals I1 and I2 and the compressibility expression C1 are functions of m and the
packing fraction n (or 𝜁3 ), while n is the reduced density.
The dispersion term is extended to mixtures by assuming the van der Waals one-fluid theory
approximation:
( )
∑∑ 𝜀ij
m2 𝜀𝜎 3 = x i x j mi mj 𝜎ij3 (3.17)
i j
kT
( )2
∑∑ 𝜀ij
m2 𝜀 2 𝜎 3 = x i x j mi mj 𝜎ij3 (3.18)
i j
kT

The parameters for a pair of unlike segments are obtained by the conventional Berthelot-
Lorentz combining rules:
1
σij = (σ + σj ) (3.19)
2 i
√
εij = εi εj (1 − kij ) (3.20)
 The binary interaction parameter, kij , is introduced to correct the segment-segment interac-
tions of unlike chains. Actually, it is a binary adjustable parameter that is determined by fitting
binary phase equilibrium data, similarly to cubic EoS.
The association contribution (aassoc ) is calculated from the following expressions:

aassoc ∑
M
= (lnX A − 0.5X A ) + 0.5M (3.21)
RT A=1

where M is the number of association sites per molecule and X A is the mole fraction of
molecules not bonded at site A. In Eq. 3.21, the summation is taken over all association sites of
the molecule. The quantity X A is calculated from the expression:
( )−1
∑
X A = 1 + NAv 𝜌X B 𝛥AB (3.22)
B

The summation in the above equation is taken over all sites. NAv is Avogadro’s number, 𝜌 is
the molar density and 𝛥AB is the association strength that is calculated from:
( )
1 − 0.5n 𝜀AB
𝛥AB = exp − 1 (𝜎 3 𝜅 AB ) (3.23)
(1 − n)3 kT
where A and B are association sites on the molecule.
The extension of the associating term to mixtures is straightforward, and no mixing rules are
required; yet, for the cross-associating compounds the following combining rules proposed by
Wolbach and Sandler [46] are used:
1
εAi Bj = (εAi Bi + εAj Bj ) (3.24)
2
( √ )3
√ σii σjj
Ai Bj A B A B
κ = κ i iκ j j
1
(3.25)
(σ + σjj )
2 ii

PC-SAFT requires three parameters for pure non-associating compounds: the segment num-
ber (m), the segment diameter (σ), and the segment energy parameter (ε/k). For pure associating
compounds, like water, alcohols, etc., two more pure-component parameters are required: the
association energy εAB and the effective associating volume κ AB . These three or five parameters
for PC-SAFT are usually determined by simultaneously fitting pure compound vapor pres-
sure and saturated liquid volume experimental data. The pure component parameters for the
non-associating components were taken from Ting et al. [35] and Gross and Sadowski [40, 41].
For water and methanol, the two-site association scheme has been adopted for both molecules
as originally proposed by Gross and Sadowski using the pure compound parameters given in
the pertinent publication [47].

3.2.3 UMR-PRU
The UMR-PRU model [39] belongs to the EoS/GE class, and couples the PR EoS [15] with
UNIFAC [48]. The temperature-dependent part of the attractive term parameter for pure
non-polar compounds is calculated from Eqs. 3.4 and 3.5, while for polar compounds by the
Mathias-Copeman [14] modification of the PR EoS:

r ) + c2 (1 − Tr ) + c3 (1 − Tr ) ] for T ≤ Tr
a(T) = [1 + c1 (1 − T0.5 0.5 2 0.5 3 2
(3.26)

r )] for T > Tr
a(T) = [1 + c1 (1 − T0.5 2
 The parameters c1 , c2 , and c3 in Eq. 3.26 are pure-component specific adjustable parameters,
which are determined by fitting pure-component vapor pressure data.
For the extension to mixtures, the following universal mixing rules (UMR) proposed by Vout-
sas et al. [32] are applied:

a
E,SG
1 GAC + GAC
E,res
∑ a
= + xi i (3.27)
bRT −0.53 RT i
bi RT
2
∑∑ ⎛ b1∕2 + b1∕2 ⎞
xi xj bij with bij = ⎜ ⎟
i j
b= (3.28)
⎜ 2 ⎟
i j ⎝ ⎠
The Staverman-Guggenheim term of the combinatorial part (GE,SG AC
) and the residual part
(GE,res
AC
) of the excess Gibbs energy (GE
) respectively, are calculated from Eqs. 3.29 – 3.32.
GE,SG
AC
∑ θ GE,res
AC
∑
xi qi ln i ,
=5 = xi vki (lnΓk − lnΓik ) (3.29)
RT φ RT
[i ( i
) i ]
∑ ∑ θm Ψmk
lnΓk = Qk 1 − ln θm Ψmk − ∑ (3.30)
m m n θn Ψnm

xr xq
For compound i∶ φi = ∑ i i θi = ∑ i i , (3.31)
j xj rj j xj qj

∑ (j)
X m Qm j νm xj
For group m∶ θm = ∑ , Xm = ∑ ∑ (j) , (3.32)
n X n Qn νn xj j n

The interaction parameter Ψmk between groups m and k (Eq. 3.30) is a function of tempera-
ture, and it is calculated by Eq. 3.33:
[ ]
Amk + Bmk (T − 298.15) + Cmk (T − 298.15)2
Ψmk = exp − (3.33)
T
where Amk , Bmk and Cmk are binary adjustable interaction parameters, which are determined
by fitting binary phase equilibrium data.
The UMR-PRU model has been successfully applied to the prediction of low- and
high-pressure vapor-liquid, liquid-liquid, solid-gas, and solid-liquid-gas equilibria as well as
heats of mixing, in binary and multicomponent systems of various degrees of non-ideality
and asymmetry with respect to the size of the molecules involved, up to those observed in
polymer/solvent systems [32, 39, 49]. Furthermore, UMR-PRU has been proven to be a very
reliable model for thermodynamic property calculations in natural gas mixtures [30, 33, 38].
The critical properties (Tc and Pc ) and the acentric factors (𝜔) are taken from DIPPR [42] as
in PR, while the Mathias-Copeman parameters c1 , c2 and c3 for the polar components are taken
by Boukouvalas et al. [50].