Fluid Phase Equilibria 307 (2011) 45–57

Contents lists available at ScienceDirect

Fluid Phase Equilibria

journal homepage: www.elsevier.com/locate/fluid

Combined chemical and phase equilibrium for the hydration of ethylene to

ethanol calculated by means of the Peng–Robinson–Stryjek–Vera equation of
state and the Wong–Sandler mixing rules
Mario Llano-Restrepo ∗ , Y. Mauricio Muñoz-Muñoz
School of Chemical Engineering, Universidad del Valle, Ciudad Universitaria Melendez, Building 336, Apartado 25360, Cali, Colombia

a r t i c l e i n f o a b s t r a c t

Article history: Due to the economics of the ethylene market and the subsidized production of fermentation-based
Received 14 February 2011 ethanol in some countries, use of the ethylene hydration process to make ethanol has been steadily
Received in revised form 9 May 2011 declining. The economics of this process might improve by combining the reaction and separation in a
Accepted 14 May 2011
reactive distillation column, whose conceptual design requires a study of the combined chemical and
Available online 23 May 2011
phase equilibrium (CPE) of the reacting system. In this work, the Peng–Robinson–Stryjek–Vera equation
of state was combined with the UNIQUAC activity coefficient model through the Wong–Sandler (WS)
Keywords:
mixing rules in order to correlate the available experimental data for the vapor–liquid equilibria (VLE)
Synthetic ethanol
Petrochemical ethanol
of the ethylene–water, ethylene–ethanol, and ethanol–water binary systems at 200 ◦ C. The interaction
Fuel ethanol energies of the UNIQUAC model and the binary interaction coefficient of the WS mixing rules were used
Ethylene hydration as the fitting parameters. From the optimum values of these parameters, both the VLE and the combined
Chemical equilibria CPE of the ethylene–water–ethanol ternary system were predicted at 200 ◦ C and various pressures. At this
Vapor–liquid equilibria temperature, the catalytic activity of a H-pentasil zeolite has already been reported to exhibit a maximum
for ethylene hydration, and also the experimentally measured two-phase region of the ternary system
is sufficiently wide. By means of the reactive flash method, the chemical equilibrium compositions of
the liquid and vapor phases were determined for several pressures, and the equilibrium conversion and
the vapor fraction were calculated as functions of the ethylene to water feed mole ratio. It turns out that
the vapor–liquid mixed-phase hydration of ethylene achieves equilibrium conversions much higher than
those computed for a vapor-phase reaction that would hypothetically occur at the same conditions of
pressure and feed mole ratio. It was found that the reactive phase diagram of the ternary system exhibits
a critical point at 200 ◦ C and 155 atm.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction combination of high ethylene price and subsidized production of

fermentation-based ethanol, use of the ethylene hydration process
1.1. Background to produce ethanol has been steadily declining in some countries,
e.g. the U.S.A., where only a single producer, Equistar Chemicals (a
In the petrochemical industry, ethanol (C2 H5 OH) is produced subsidiary of LyondellBasell Industries), remains in the field with
from the catalytic hydration of ethylene (C2 H4 ), according to the its plant (with a capacity of 150 kt/year) at Tuscola, Illinois [2–5].
reaction: In contrast, synthetic ethanol is made at low cost in places where
ethylene is cheap, e.g. Saudi Arabia, where 300 kt/year of crude
C2 H4 + H2 O ↔ C2 H5 OH ethanol are produced at Al-Jubail, by SADAF, a subsidiary of SABIC
[1,3,6]. Europe’s largest producers of synthetic ethanol are INEOS
Ethanol produced from ethylene is known as synthetic (or petro- Enterprises with two plants at Grangemouth, Scotland (with a total
chemical) ethanol. Synthetic ethanol is still used as an intermediate capacity of 310 kt/year) [7,8], and Sasol Solvents Germany, with one
in the manufacture of other chemicals such as ethyl acrylate, ethyl plant at Herne, Germany, with a production capacity of 140 kt/year
acetate, ethylamines, diethyl ether, ethyl vinyl ether, ethyl tert- [9].
butyl ether, and glycol ethyl ethers [1]. Due to the unfavorable The vapor-phase direct hydration of ethylene, using phosphoric
acid as catalyst, has been employed for the commercial production
of synthetic ethanol since 1948 [10,11]. A calcined diatomaceous
∗ Corresponding author. Tel.: +57 2 3312935; fax: +57 2 3392335. earth known as Celite used to be the preferred catalyst support
E-mail address: [email protected] (M. Llano-Restrepo). until a porous silica xerogel was adopted [12]. The process of direct

0378-3812/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.fluid.2011.05.007
46 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57

Table 1
List of works in which experimental measurements of high-pressure VLE have been reported for the ethylene–water, ethylene–ethanol, and ethanol–water binary systems,
and the ethylene–water–ethanol ternary system.

System Published work Temperature Pressure

range (◦ C) range (atm)

Ethylene–water Tsiklis et al. [33] 200–300 30–160

Anthony and McKetta [34] 38–138 14–340
Sanchez and Lentz [35] 150–300 99–691

Ethylene–ethanol Tsiklis and Kofman [36] 150–220 10–127

Ethanol–water Griswold et al. [37] 150–275 7–102

Barr-David and Dodge [38] 150–350 5–187

Ethylene–water–ethanol Tsiklis et al. [39] 200–300 30–160

hydration of ethylene consists of three sections: reaction, recov- [36], and ethylene–water–ethanol [39]. Table 2 provides a list of
ery and purification, and production of anhydrous ethanol. Brief works [40–51] in which equations of state (EoS) combined with
descriptions of the process have been given in Refs. [1,13–17]. excess Gibbs energy (GE ) models, through particular sets of mixing
Detailed descriptions can be found in Refs. [11,18,19]. The ethy- rules, have already been used to correlate (or predict) the VLE data
lene to water mole ratio in the reactor feed is in the range of reported by Barr-David and Dodge [38] for ethanol–water.
0.3:1 to 4:1. Reaction temperatures are in the range of 210–325 ◦ C. Table 3 provides a list of works [52–58] in which chemical
Reactor pressures are in the range of 30–100 atm. Use of high tem- equilibrium calculations for the vapor-phase direct hydration of
peratures increases the reaction rate; however, since the reaction ethylene have been reported. The temperature-dependency for the
is exothermic [1,17–19], its equilibrium conversion decreases at chemical equilibrium constant used in all these works was based
higher temperatures, and the reaction is then limited by chem- on values of the standard properties of formation that turn out to
ical equilibrium. In consequence, the temperature of the reactor be inaccurate by comparison with those more recently determined
should be set as a compromise between kinetics and thermody- [59,60]. Ideal-solution behavior for the vapor phase was assumed
namics. The gas–liquid mixed-phase hydration of ethylene, feeding in Refs. [52–54].
gaseous ethylene and liquid water into a trickle-bed reactor [19], Table 4 provides a list of works in which combined chemical and
has been the subject of several patents [20–24] that use a mixture phase equilibrium (CPE) calculations for the vapor–liquid mixed-
of titanium and tungsten oxides [20,22], the blue oxide of tungsten phase hydration of ethylene have been reported. These calculations
[21,23], or a dispersed alumina gel [24] as the catalyst. The water to were based on some simplifying assumptions. The solubility of
ethylene mole ratio in the reactor feed is in the range of 16:1 to 54:1. ethylene in the liquid phase was neglected in all three works and
Reaction temperatures are in the range of 240–310 ◦ C. Reactor pres- use of the gamma-phi formulation was made in Refs. [54,62] with-
sures are in the range of 100–350 atm. However, as far as is known out applying any pressure-correction for the liquid-phase activity
[19], none of these patents led to any commercial applications. coefficients of ethanol and water, which were estimated in Ref. [62]
Some technology enhancements for the vapor-phase direct from the isobaric VLE data of Otsuki and Williams [63] and corre-
hydration of ethylene have been proposed and described recently lated (by means of the van Laar model) in Ref. [54] from the VLE
[25]. According to their proponents [25–27], these enhancements
can improve the economics of the process in such an extent that Table 2
they can make synthetic ethanol competitive with fermentation- List of works in which equations of state (EoS) have been combined with excess Gibbs
based ethanol due to a simpler plant design, lower construction and energy (GE ) models to correlate (or predict) the high-pressure VLE data measured
operating costs, and a higher purity product [26]. The economics of by Barr-David and Dodge [38] for the ethanol–water binary system.
the direct hydration of ethylene might improve further by applying Published work EoS GE model Mixing rules
the concept of process intensification [28,29]. In an intensified pro-
Gupte et al. [40] vdW UNIFAC-Lingby VD
cess, the reaction and the separation of the product (ethanol) and Schwartzentruber and MSRK UNIFAC NQCD
the reactants (ethylene and water) would occur in a single piece Renon [41]
of equipment, a reactive distillation (RD) column. Indeed, ethy- Michelsen [42] SRK Wilson IVI
lene hydration has been mentioned [30] as one of several potential Dahl and Michelsen [43] MCSRK UNIQUAC MHV2
Wong et al. [44] PRSV van Laar WS
applications of RD in the presence of a supercritical fluid (ethylene, Fischer and Gmehling MCSRK UNIFAC HFG
in this case). A study of the combined chemical and phase equi- [45]
librium (CPE) of the reacting system is required for the conceptual Hernández-Garduza et al. PRCRP van Laar RPSF
design of an RD column [31,32]. [46]
res
Constantinescu et al. [47] MCSRK UNIQUAC HVID
Smith et al. [48] PR van Laar SVA
1.2. Prior work Voutsas et al. [49] MTPR UNIFAC-Lingby UMR
Shimoyama et al. [50] SKYSRK COSMO-RS MHV1
Table 1 provides a list of works [33–39] in which experimental Lee and Lin [51] PR COSMO-SACres WS

measurements of VLE at high pressures have been reported for the vdW: van der Waals; SRK: Soave–Redlich–Kwong; MSRK: Mathias-SRK; MCSRK:
ethylene–water, ethylene–ethanol, and ethanol–water binary sys- Mathias–Copeman-SRK; SKYSRK: Sandarushi–Kidnay–Yesavage-SRK; PR:
Peng–Robinson; PRSV: PR-Stryjek–Vera; PRCRP: PR-Carrier–Rogalski–Peneloux;
tems, and the ethylene–water–ethanol ternary system. From the res
MTPR: Magoulas–Tassios-PR; UNIQUAC : universal quasichemical residual
consideration of the typical temperature and pressure ranges for contribution; COSMO-RS: conductor-like screening model for real solvents;
the ethylene hydration reaction (see Section 1.1) or the number of COSMO-SACres : COSMO-segment activity coefficient residual contribution;
data points measured, the VLE data reported by Tsiklis et al. [33] for VD: volume-dependent; NQCD: non-quadratic composition-dependent; IVI:
ethylene–water and Barr-David and Dodge [38] for ethanol–water implicit volume-independent; MHV2: modified Huron–Vidal second order; WS:
Wong–Sandler; HFG: Holderbaum–Fischer–Gmehling; RPSF: reference packing
were chosen for the present work. fraction and scaling factor; HVID: Huron–Vidal with infinite dilution coefficient,
No previous attempts have been made to correlate the VLE data SVA: Smith–Van Ness–Abbott; UMR: universal mixing rule; MHV1: modified
reported by Tsiklis et al. for ethylene–water [33], ethylene–ethanol Huron–Vidal first order.
 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57 47

Table 3
List of works in which chemical equilibrium calculations for the vapor-phase hydration of ethylene to ethanol have been reported.

Published work Temperature (◦ C) Pressure (atm) C2 H4 /H2 O feed mole ratio Fugacity coefficient calculation
method

Ewell [52] 150 5–200 0.06–41.6 Newton chart + Berthelot EoS

Vvedensky and Feldman [53] 150–400 50–200 0.2–5.0 Newton–Dodge chart
Wenner [54] 300 80 0.25, 1.0 Dodge chart
Muller and Waterman [55] 200–325 20–100 1.0–4.0 Berthelot EoS
Beattie–Bridgeman EoS
Redlich–Kwong EoS
Bakshi et al. [56,57] 250–365 1–100 0.32–3.0 Truncated virial EoS
Tsiklis and Kulikova [58] 200–300 10–100 – Partial molar volumes

Table 4
List of works in which combined chemical and phase equilibrium calculations for the vapor–liquid mixed-phase hydration of ethylene to ethanol have been reported.

Published work Temperature (◦ C) Pressure (atm) Fugacity calculation method

Liquid-phase Vapor-phase

Wenner [54] 250 80 Van Laar model Hougen–Watson charts

Cope and Dodge [61] 200, 250 100, 200, 300 Partial molar volumes Hougen–Watson charts
Smith et al. [62] 200 34 Experimental data Truncated virial EoS

data reported by Griswold et al. [37]. In Ref. [61], the liquid-phase the ternary system are presented, discussed and compared with the
partial fugacities for ethanol and water were estimated from their available experimental data [39]. Lastly, in Section 4, model predic-
saturated fugacities (obtained from the data of Griswold et al. [37]) tions for the combined CPE of the reacting system are presented,
by making some assumptions on the partial molar volumes of water discussed and compared with those previously reported [61,62].
and ethanol in order to include the pressure effect on the fugaci-
ties. In all three works, the partial fugacity coefficients in the vapor
2. Correlation of binary VLE data
phase were computed from the assumption of ideal solution, by
means of the generalized fugacity charts of Hougen and Watson
The model used in the present work is comprised of the
[64] in Refs. [54,61], and the truncated virial equation of state (using
Peng–Robinson–Stryjek–Vera (PRSV2) equation of state, the UNI-
the Pitzer–Curl–Abbott correlation for the second virial coefficient
QUAC activity coefficient model, and the Wong–Sandler (WS)
of non-polar gases) in Ref. [62]. Inaccurate values of the chemical
mixing rules. The reader is referred to Appendix A for a complete
equilibrium constant were used in Refs. [54,61].
description of the model. The basic parameters required for the
application of the PRSV2-EoS and the UNIQUAC model are given
1.3. Overview in Tables 5 and 6. Table 7 describes the type, source, and number
of data points involved, and the calculation algorithm and opti-
In this work, the Peng–Robinson equation of state [65] with mization method used in the regression of parameters for the fit of
the Stryjek–Vera [66] three-adjustable-parameter form for the the PRSV2 + UNIQUAC + WS model to the experimental VLE data of
attractive term (PRSV2-EoS) is combined with the UNIQUAC model ethylene–water, ethylene–ethanol, and ethanol–water.
[67] through the Wong–Sandler (WS) mixing rules [68] in order Flow charts for T–P flash calculations (using equations of state)
to correlate the available experimental data for the VLE of the have been given in some thermodynamics textbooks [82,83]. The
ethylene–water [33], ethylene–ethanol [36], and ethanol–water (L) (V )
equilibrium condition yi /xi = ϕ̂i /ϕ̂i ≡ Ki is combined with the
[38] binary systems at 200 ◦ C. The interaction energies of the UNI- component mole balance equation zi F = xi L + yi (F − L) and the rela-
QUAC model and the binary interaction coefficient of the WS mixing N
tion (x − yi ) = 0 to yield the Rachford–Rice equation [84]:
i=1 i
rules are used as the fitting parameters. From the optimum values
of these parameters, both the VLE of the ethylene–water–ethanol 
N
ternary system and the combined CPE of the reacting system are zi (1 − Ki )
=0 (1)
predicted at 200 ◦ C and various pressures. At this temperature, the Ki + L (1 − Ki )
i=1
catalytic activity of a H-pentasil zeolite has already been reported
[69] to exhibit a maximum for ethylene hydration, and also the In this work, Eq. (1) was solved for the liquid fraction  L = L/F by
experimentally measured [39] two-phase region of the ternary sys- means of the Newton–Raphson method, calculating the derivative
tem is sufficiently wide. The combined CPE is calculated by means with respect to  L analytically from Eq. (1). The partial fugacity
of the reactive flash method. The outline of the paper is as follows. coefficients in both phases were calculated from Eqs. (A.13)–(A.17)
The PRSV2-EoS + UNIQUAC + WS model is described in Appendix with the help of Eqs. (A.7)–(A.12) and (A.18)–(A.21). That the liquid
A. In Section 2, the method used for the correlation of the avail- and vapor compositions resulting from the application of this algo-
able VLE data for the binary systems is explained and results of the rithm are indeed true equilibrium values was checked by means
model fitting are presented. In Section 3, predictions for the VLE of of the tangent–plane distance criterion, from the decrease in the

Table 5
Values of the critical temperature Tc , critical pressure Pc , acentric factor ω, and parameters 1 , 2 , 3 to be used for the calculation of the attractive term of the PRSV2 equation
of state.

Component Tc (K) Pc (bar) ω 1 2 3 Ref.

Ethylene 282.35 50.42 0.08652 0.04191 0.0 0.0 [71]

Water 647.286 220.8975 0.34380 −0.06635 0.0199 0.443 [66,70]
Ethanol 513.92 61.48 0.64439 −0.03374 −2.6846 0.592 [66,70]
48 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57

Table 6 N
bined with the stoichiometric constraint y
i=1 i
= 1 to yield the
Values (taken from Ref. [81]) of the molecular volume (ri ) and surface area (qi )
parameters used for the application of the UNIQUAC model. following equation:
 
Component ri qi 
N
f (P) = ln Ki xi =0 (3)
Ethylene 1.57 1.49
Water 0.92 1.40 i=1
Ethanol 2.11 1.97
Eq. (3) was solved for the implicit variable P by means of
the Newton–Raphson method, calculating the derivative f (P) by
numerical perturbation. P–y bubble-point calculations were carried
dimensionless Gibbs free energy change [85,86]. T–P flash calcula- out to correlate the VLE data of ethanol–water. To obtain the opti-
tions were carried out to correlate the VLE data of ethylene–water mum values of the fitting parameters, the Nelder and Mead method
and ethylene–ethanol. To obtain the optimum values of the fitting [87] was used to minimize the following objective function:
parameters, either the flexible polyhedron method for multidimen-  2  2 

nd
y3,calc − y3,exp Pcalc − Pexp
sional optimization by Nelder and Mead [87] or a random search Fobj = + (4)
optimization algorithm [88] were used to minimize the following y3,exp Pexp
i=1
objective function:
The optimum values of the fitting parameters and the correspond-
 2  2 

nd
x1,calc − x1,exp y1,calc − y1,exp
ing deviations are reported in Table 8. The resulting fit is shown in
Fobj = + (2) Fig. 3. Table 9 provides a comparison between the deviations associ-
x1,exp y1,exp
i=1 ated to this fit and those corresponding to the fits reported by other
authors in previous works [40,43,45–51], in which an equation of
The optimum values of the fitting parameters and the correspond- state had also been combined with an excess Gibbs energy model
ing deviations are reported in Table 8. The resulting fits are shown through a set of mixing rules (as described in Table 2). It follows
in Figs. 1 and 2. If one considers that the second virial coeffi- that our fit is the most accurate one so far reported (its accuracy
cients resulting from the PR-EoS are only approximate and the exceeding by far that of most the other works).
combining rule for the cross second virial coefficient, Eq. (A.11),
is just an empirical approximation, the resulting value of 1.0181 3. Prediction of ternary VLE data
for k12 is not abnormal. By using the WS mixing rules to fit VLE
data (at various temperatures), Huang et al. [89] reported values As the optimum values of the adjustable parameters for the
of kij from 1.01 to 1.15 for binary systems of hydrogen with n- PRSV2 + UNIQUAC + WS model have already been determined for
butane, n-hexane, n-heptane, n-octane, n-decane, n-hexadecane, the three binary systems at 200 ◦ C (see Table 8), it is possi-
and toluene, and López et al. [77] reported a value of 1.02 for the ble to make predictions from the model for the VLE of the
systems nitrogen–dimethyl ether and helium–toluene, and values ethylene–water–ethanol ternary system at 200 ◦ C and compare the
of 1.43 (at 35 ◦ C) and 2.42 (at 45 ◦ C) for nitrogen–methanol. results with the corresponding experimental data of Tsiklis et al.
Flow charts for P–y bubble-point calculations (using equations [39]. The T–P flash algorithm described in Section 2 was also used to
of state) have been given in some thermodynamics textbooks calculate the VLE of the ternary system. That the resulting liquid and
[82,83,90,91]. In this work, the equilibrium condition was com- vapor compositions are indeed true equilibrium values was also

160

140

120
Liquid
Pressure (atm)

100

80

Vapor
60

40

20
0.000 0.002 0.004 0.006 0.008 0.45 0.55 0.65 0.75 0.85 0.95
mole fraction of ethylene

Fig. 1. Fit of the PRSV2 + UNIQUAC + WS model to the Pxy phase diagram of the ethylene–water binary system at 200 ◦ C. (䊉) Experimental data of Tsiklis et al. [33]; (—)
calculated with the model.
 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57 49

Table 7
Description of the data and calculation method used for the fit of the PRSV2 + UNIQUAC + WS model to the experimental VLE data of the ethylene (1) + water (2), ethylene
(1) + ethanol (3), and ethanol (3) + water (2) binary systems at 200 ◦ C.

System (ij) No. of Type of data Data source Calculation Optimization Objective
data points used algorithm method function

12 9 P, x1 , y1 [33] T, P flash Nelder–Mead Eq. (2)

13 6 P, x1 , y1 [36] T, P flash Random search Eq. (2)
32 17 P, x3 , y3 [38] P, y bubble Nelder–Mead Eq. (4)

Table 8
Optimum values of the adjustable parameters and deviations for the fit of the PRSV2 + UNIQUAC + WS model to the experimental VLE data of the ethylene (1) + water (2),
ethylene (1) + ethanol (3), and ethanol (3) + water (2) binary systems at 200 ◦ C.

System (ij) uij /R (K) uji /R (K) kij x1 a y1 b y3 b P/Pc (%)
−5
12 808.94 385.66 1.0181 4 × 10 0.0056 – –
13 420.00 100.00 0.2600 0.0052 0.0162 – –
32 23.67 369.45 0.0085 – – 0.0032 1.10

 1 

nd

a
x = n

xkcalc − xkexp
.
d

k=1

 

nd

b
y = n1
ykcalc − ykexp
.
d

k=1
nd

 
Pkcalc −Pkexp
.
c P
P
= 100
nd
Pexp

k
k=1

Table 9
Comparison of deviations for the vapor-phase mole fraction of ethanol (y) and the mixture bubble-point pressure (P) with respect to the experimental data [38] for the
ethanol–water binary system at 200 ◦ C.

Published work ya P/P (%)b y/y (%)c RMSQD(y)d RMSQD(P) (%)e ı(y, P) (%)f

Gupte et al. [40] 0.0196 2.49

Dahl and Michelsen [43] 0.0040 0.90
Fischer and Gmehling [45] 0.0080 2.40
Hernández-Garduza et al. [46] 0.0100 1.39
Constantinescu et al. [47] 0.0110 1.60
Smith et al. [48] 0.0050 1.40
Voutsas et al. [49] 0.0170 2.00
Shimoyama et al. [50] 3.38
Lee and Lin [51] 3.05 3.89
This work 0.0032 1.10 0.82 0.0038 1.31 0.71

 1 
calc exp

nd

a
y = n

yk − yk
.
d

k=1

 
calc exp

n

Pk −Pk
.
d

b P
P
= 100
nd
Pexp

k
k=1

 

ykcalc −ykexp

nd

c y
y
= 100
nd
yexp
.
k

k=1 1/2
 1 
nd

exp 2
d
RMSQD(y) = nd
(ykcalc − yk ) .

 nd 
k=1

 1  P calc −P
exp
2 1/2
e
RMSQD(P) = 100 nd
k k
exp .
P
k

 n nd

k=1

 50 



exp

exp
P calc −P
f
ı(y, P) = n
d
yk − yk
calc
+
exp
.
k k
P
k
k=1 k=1

Table 10
Pure component gas-phase standard properties of formation (kJ/mol) at 25 ◦ C [59,60] and coefficients [94,95] of the ideal-gas heat capacity correlation, Eq. (9).

Component Hf0 Gf0 a b × 103 c × 106 d × 10−5

Ethylene 52.51 68.46 1.424 14.394 −4.392 0

Water −241.818 −228.572 3.470 1.450 0 0.121
Ethanol −235.10 −168.49 3.518 20.001 −6.002 0
50 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57

100 and (E). At a temperature of 200 ◦ C, the critical point of a mixture of

ethylene and ethanol occurs at a pressure close to 100 atm [36,39];
90 therefore, in the phase diagram that would correspond to that pres-
sure, the dew-point and bubble-point lines of the ternary system
80 intersect the ethylene–ethanol side of the diagram at the compo-
Liquid sition of that critical point. In Fig. 4(F), at a pressure of 100 atm,
the intersection of the two lines appears to be imminent. For pres-
Pressure (atm)

70
sures above 100 atm, all mixtures of ethylene and ethanol are single
Vapor
60 phase, only the ethylene–water binary exhibits a two-phase region,
and the predicted dew-point and bubble-point lines of the ternary
50 system join at a critical point, where the two phases become iden-
tical.
40
4. Combined chemical and phase equilibrium
30
4.1. Chemical equilibrium constant
20
0.0 0.1 0.2 0.3 0.4 0.5 Since gas-phase thermochemical quantities are more readily
mole fraction of ethylene found, it is preferable to formulate the chemical equilibrium con-
dition for the vapor phase, as follows [92,93]:
Fig. 2. Fit of the PRSV2 + UNIQUAC + WS model to the Pxy phase diagram of the
ethylene–ethanol binary system at 200 ◦ C. (䊉) Experimental data of Tsiklis and 
N  P −
(V ) i
Kofman [36]; (—) calculated with the model. [yi ϕ̂i ] = K (5)
P0
i=1

checked by means of the tangent–plane distance criterion, from the The chemical equilibrium constant K depends only on the temper-
decrease in the dimensionless Gibbs free energy change [85,86]. ature through the following expression:
The resulting predictions at various pressures are reported in  
0
Grx
the ternary phase diagrams shown in Fig. 4. In contrast to the
K = exp − (6)
paucity of experimental data points in those diagrams, the model RT
predictions extend over larger ranges. The upper locus (indicated
by the triangles) of the ternary diagram is the dew-point line, K can be obtained by integration of the van’t Hoff’s equation:
the lower locus (indicated by the circles) is the bubble-point  0
Hrx
line, and the region within the saturation envelope corresponds ln K = dT + J (7)
to the two-phase region. For pressures above 29 atm (see Fig. 3) RT 2
ethanol–water mixtures are all liquid. For pressures below 100 atm, 0 is in turn obtained by integration of Kirchhoff’s equa-
where Hrx
both the ethylene–ethanol and the ethylene–water binaries exhibit tion:
two-phase regions, with their bubble-point and dew-point compo- 
sitions appearing along the corresponding sides of the diagram as 0 0
Hrx = Cp,rx dT + I (8)
the ends of the bubble-point and dew-point lines of the ternary
system. The excellent fit of the bubble-point and dew-point com-
positions of the two binary systems can be appreciated in Fig. 4(B) Cp0 can be obtained from the following correlation [92]:

CP0
= a + bT + cT 2 + dT −2 (9)
31 R
The integration constants I and J are evaluated by applying Eqs.
29

N

N
(6)–(9) at 25 ◦ C, where Hrx
0 = 0 and G0 =
i Hf,i rx
0 .
i Gf,i
27
Liquid i=1 i=1
When the values (readily found in Ref. [92]) given in Table 10 for the
Pressure (atm)

25 standard properties of formation at 25 ◦ C (taken from Ref. [59] for

ethylene and Ref. [60] for water and ethanol), and the coefficients
23 a, b, c, d of Eq. (9) (taken from Ref. [94] for ethylene and ethanol,
Vapor
and Ref. [95] for water), are used to carry out the integrations indi-
21 cated in Eqs. (7) and (8), and I and J are determined, the following
expression is obtained:
19
ln K = −1.376 ln T + 2.0785 × 10−3 T − 2.683 × 10−7 T 2
17 6050 5308.358
− + − 7.1126 (10)
T2 T
15
where P0 = 1 bar in Eq. (5). At 200 ◦ C, K = 0.031042. Since the values
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
of the standard properties of formation on which Eq. (10) is based
mole fraction of ethanol
are more accurate than those associated with other expressions for
Fig. 3. Fit of the PRSV2 + UNIQUAC + WS model to the Pxy phase diagram of the
K available in the literature [52–55,96–102], chemical equilibrium
ethanol–water binary system at 200 ◦ C. (䊉) Experimental data of Barr-David and calculations using Eq. (10) are expected to be more reliable than
Dodge [38]; (—) calculated with the model. those reported in Refs. [52–58] for the vapor-phase reaction.
 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57 51

Fig. 4. Phase diagram of the ethylene–water–ethanol ternary system at 200 ◦ C and various pressures: (A) 30 atm, (B) 40 atm, (C) 50 atm, (D) 60 atm, (E) 80 atm, (F) 100 atm,
(G) 120 atm, (H) 140 atm, and (I) 160 atm. Predictions from the PRSV2 + UNIQUAC + WS model: () vapor phase; () liquid phase. Experimental data of Tsiklis et al. [33,36,39]:
() vapor phase; (䊉), liquid phase.

4.2. Equilibrium phase compositions For the phi–phi formulation of the reactive flash method, the
phase equilibrium condition is given by the expression Ki ≡ yi /xi =
Algorithms for combined CPE calculations in multiple-reaction (L) (V )
ϕ̂i /ϕ̂i . From Eq. (12), it follows that L = n − V. Substitution of this
systems have been given by Sanderson and Chien [103], and Xiao expression for L and xi = yi /Ki into Eq. (11), and use of the first part
et al. [104] using the gamma-phi formulation (i.e. an activity coef- of Eq. (12) lead to:
ficient model for the liquid phase and an equation of state for the
Ki (ni,0 + i εe )
vapor phase). A flow chart for these calculations has been given by yi = (13)
O’Connell and Haile [105]. An improved algorithm that involves a (n0 + εe ) + V (Ki − 1)
three-stage overall strategy and uses the phi–phi formulation (i.e. Since the vapor-phase mole fractions sum to unity, the following
an equation of state for both phases), was proposed and tested reactive flash equation is obtained from Eq. (13):
by Stateva and Wakeham [106]. However, for the case of single-
reaction systems and the gamma–phi formulation, a much simpler

N
Ki (ni,0 + i εe )
−1=0 (14)
reactive flash algorithm has been given by Walas [107]. For the (n0 + εe ) + V (Ki − 1)
i=1
present work, this algorithm was adapted to the phi–phi formula-
tion. The chemical equilibrium condition, Eq. (5), can be rewritten in the
In terms of the equilibrium extent of reaction εe , the compo- following form:
nent and total mole balances are given by the following expressions
 P  
N
[107]: (V ) i
[yi ϕ̂i ] − K = 0 (15)
P0
i=1
ni = ni,0 + i εe = Lxi + Vyi (11)
After specifying T and P, and the ethylene to water feed mole ratio
n = n0 + εe = L + V (12) r = n1,0 ,n2,0 , and setting n1,0 = r, n2,0 = 1, and n3,0 = 0, the values of
52 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57

Table 11 a b c d e f g
AB C
Mole fractions in the liquid (x1 , x2 , x3 ) and vapor (y1 , y2 , y3 ) phases for the combined 100 D 100
E
chemical and phase equilibrium of the ethylene (1) + water (2) + ethanol (3) ternary F

system at 200 ◦ C, as predicted from the PRSV2 + UNIQUAC + WS model. 90 G 90

C2H4 conversion (%)

P (atm) x1 x2 x3 y1 y2 y3 80 80

30 0.0010 0.9425 0.0565 0.3050 0.5150 0.1800 70 70

vapor (%)
40 0.0027 0.8742 0.1231 0.3835 0.3816 0.2349
50 0.0071 0.7625 0.2304 0.4314 0.2974 0.2712 60 60
60 0.0169 0.6330 0.3501 0.4662 0.2377 0.2961
70 0.0315 0.5263 0.4422 0.4920 0.1945 0.3135 50 50
80 0.0497 0.4445 0.5058 0.5096 0.1635 0.3269
40 40
90 0.0707 0.3807 0.5486 0.5200 0.1413 0.3387
100 0.0945 0.3292 0.5763 0.5237 0.1256 0.3507 30 30
110 0.1210 0.2865 0.5925 0.5213 0.1147 0.3640
120 0.1509 0.2499 0.5992 0.5128 0.1076 0.3796 20 20
130 0.1852 0.2175 0.5973 0.4973 0.1040 0.3987
140 0.2269 0.1872 0.5859 0.4724 0.1040 0.4236 10 10
150 0.2858 0.1554 0.5588 0.4284 0.1100 0.4616
0 0
154 0.3297 0.1375 0.5328 0.3900 0.1188 0.4912
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
C2H4 / H2O feed ratio

V and εe , (and, in consequence, those of xi and yi ) are obtained Fig. 5. Percentage conversion of ethylene (solid lines, with upper case labels) and
from the simultaneous solution of Eqs. (14) and (15) by means of vapor percentage (dashed lines, with lower case labels) as functions of the ethylene
to water feed mole ratio and pressure for the combined chemical and phase equilib-
the Newton–Raphson method, with the help of Eq. (13) and Eqs.
rium at 200 ◦ C. From left to right the curves correspond to the following pressures:
(A.7)–(A.21). (A, a) 30 atm, (B, b) 40 atm, (C, c) 50 atm, (D, d) 60 atm, (E, e) 70 atm, (F, f) 80 atm,
With the optimum values of the parameters given in Table 8, the and (G, g) 100 atm.
PRSV2 + UNIQUAC + WS model was used to predict the combined
chemical and VLE of the reacting ternary system. The resulting
4.3. Equilibrium conversion and vapor fraction
liquid and vapor-phase compositions are reported in Table 11. To
make sure that the compositions resulting from the reactive flash
The number of degrees of freedom F to fix the intensive state
algorithm are indeed chemical and phase equilibrium values, the
of a multi-phase reacting system with N components is given by
algorithm was run for several values of the ethylene to water feed
the phase rule, F = N −  + 2 − Rx , where  is the number of phases
mole ratio r and the expected invariance of the final values of the
and Rx is the number of linearly independent chemical reactions
mole fractions {xi } and {yi } with respect to r (see Section 4.3) was
[108]. To study the combined CPE for the hydration of ethylene to
confirmed. The mole fractions of ethylene (x1 and y1 ) increase with
ethanol, there are two degrees of freedom to fix the intensive state
increasing pressure, except for pressures above 110 atm, for which
of the ethylene–water–ethanol ternary system because N = 3,  = 2,
a small decrease occurs for y1 . The mole fractions of ethanol (x3
and Rx = 1. In consequence, specification of both T and P fixes the
and y3 ) also increase with increasing pressure, except for pressures
intensive state of the system, regardless of the initial amounts {ni,0 }
above 120 atm, for which a small decrease occurs for x3 .
of the components. This is equivalent to say that at fixed values of T
The ethylene (1) + water (2) + ethanol (3) reacting system is a
and P, the equilibrium compositions of the vapor and liquid phases,
highly non-ideal system at 200 ◦ C. Over the range of 30–154 atm,
(V )
{xi } and {yi }, are independent of the ethylene to water feed mole
the vapor-phase partial fugacity coefficients ϕ̂i monotonically ratio r = n1,0 /n2,0 . Thus, these compositions can be obtained from
decrease from 0.8892 to 0.3715 for water and from 0.7614 to
0.2324 for ethanol, and fall within the range of 1.0020–1.3697 for
ethylene; the liquid-phase activity coefficients (computed from
(L)
the expression
i = ϕ̂i /ϕi ) monotonically decrease from 336 100
to 1.9305 for ethylene, monotonically increase from 1.0108 to
3.1609 for water, and fall within the range of 0.9736–3.2811 for 90
ethanol.
C2H4 conversion (%)

80

Table 12
Minimum and maximum values of the ethylene to water feed mole ratio r for the 70
existence of two phases in the ethylene–water–ethanol ternary system at chemical
equilibrium at 200 ◦ C, as predicted from the PRSV2 + UNIQUAC + WS model. 60
P (atm) rmin rmax
50
30 0.0575 0.6977
40 0.1262 1.0030
50 0.2393 1.2358 40
B C D
60 0.3734 1.4279 A
70 0.4891 1.5855
80 0.5845 1.7058
30
90 0.6665 1.7887 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
100 0.7408 1.8359 C2H4 / H2O feed ratio
110 0.8118 1.8496
120 0.8834 1.8317
Fig. 6. Equilibrium percentage conversion of ethylene as a function of the ethylene
130 0.9605 1.7826
to water feed mole ratio at 200 ◦ C. Solid lines: combined chemical and phase equi-
140 1.0513 1.6985
librium; dashed lines: hypothetical vapor-phase chemical equilibrium at the same
150 1.1826 1.5568
conditions. From left to right the curves correspond to the following pressures: (A)
154 1.2867 1.4446
30 atm, (B) 40 atm, (C) 50 atm, and (D) 60 atm.
 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57 53

Table 13
Liquid and vapor-phase compositions for the combined chemical and phase equilibrium of the ethylene (1) + water (2) + ethanol (3) ternary system at 200 ◦ C, as predicted in
the present work and estimated by Cope and Dodge [61] and Smith et al. [62].

P (atm) x1 x2 x3 y1 y2 y3 Ref.

34 0.0000 0.9400 0.0600 0.3560 0.4640 0.1800 [62]

34 0.0015 0.9191 0.0794 0.3425 0.4522 0.2053 This work
100 0.0000 0.7800 0.2200 0.5280 0.2020 0.2700 [61]
100 0.0945 0.3292 0.5763 0.5237 0.1256 0.3507 This work

the reactive flash algorithm described in Section 4.2 by specifying Table 12 provides the minimum (bubble-point) and maximum
any feasible value of r (leading to final values of L and V such that (dew-point) values of r for the existence of two phases in the react-
L = n0 + εe − V ≥ 0). ing ternary system at 200 ◦ C as a function of the pressure. The
The additional specification of {ni,0 } fixes the extensive state of range of r over which the two-phase system occurs can be used to
the system, i.e. it allows the determination of the reaction extent εe determine whether the conditions chosen to calculate the chemi-
and the liquid and vapor-phase amounts L and V. Thus, the vapor cal equilibrium for the vapor-phase reaction are valid or not. In a
fraction  V = V/(L + V) in the two-phase system and the fractional previous study [55], calculations for the equilibrium conversion of
conversion of ethylene,  1 = εe /n1,0 turn out to depend on r. Once the vapor-phase reaction were carried out at several conditions of
{xi } and {yi } (at fixed values of T and P) have been determined temperature, pressure, and ethylene to water feed mole ratio. In
from the reactive flash algorithm by specifying one feasible value particular, results were reported at 200 ◦ C and pressures of 40, 50,
of r, values of  V and  1 can be calculated for any other values of r, 70, 90, and 100 atm, for r = 1 (see Table XIV in Ref. [55]). However, as
in order to determine the range of values of r over which  V spans follows from Table 12, it happens that according to the predictions
the interval [0, 1], i.e. the system is in the two-phase region. As will of the PRSV2 + UNIQUAC + WS model, the reaction would actually
be shown next, it is possible to obtain analytical expressions for the occur in a two-phase system at those conditions.
calculation of  V and  1 in terms of r and x1 , x2 , y1 , and y2 .
The second part of the component mole balance, Eq. (11), 4.4. Comparison with previous works
applied to the three components of the reacting system leads to
a set of three equations in three unknowns (εe , L, V). By setting Some calculations of the combined chemical and VLE for the
n1,0 = r, n2,0 = 1 and n3,0 = 0, it follows that n0 = 1 + r; therefore, from ethylene–water–ethanol system at 200 ◦ C have been reported in
Eq. (12), the expression εe = (1 + r) − (L + V) is obtained. The analyti- two previous works [61,62]. As explained in Section 1.2, these cal-
cal solution of the set of equations yields the following expressions culations were based on some simplifying assumptions. Table 13
for  V and  1 : provides a comparison between the liquid and vapor-phase equilib-
V rium compositions predicted by the PRSV2 + UNIQUAC + WS model
V = (16) and those estimated in Refs. [61,62]. In contrast to the assumption
L+V
made in those two previous works, the computed mole fraction
1
1 = [(1 + r) − (L + V )] (17) of ethylene in the liquid phase (x1 ) is not negligible, particularly at
r
100 atm. The smallest discrepancies with respect to our results cor-
where respond to the vapor-phase compositions calculated in Ref. [62] at
r(x2 − x1 ) − (1 − r)(1 − x2 ) 34 atm. The ideal-solution vapor-phase fugacity coefficient values
V= (18) of 0.963, 0.846, and 0.753 for ethylene, water, and ethanol, respec-
(1 − y2 )(x2 − x1 ) − (y2 − y1 )(1 − x2 )
tively, calculated by Smith et al. [62] do not deviate much from our
and partial fugacity coefficient values 1.033, 0.881, and 0.735.
(1 − r) − V (y2 − y1 )
L= (19)
x2 − x1
180
 V and  1 were calculated from Eqs. (16)–(19) as functions of r
for various values of the pressure. Results are shown in Fig. 5. As C
expected, at fixed pressure,  1 decreases as the value of r increases. 150
Also, as expected from Le Châtelier’s principle, due to the negative
change in the total number of moles by the reaction, at a given
120
Pressure (atm)

value of r,  1 increases with increasing pressure. At fixed pressure,

 V increases with increasing value of r. Also, at a given value of r, Liquid
 V decreases as the pressure increases. As follows from Fig. 5, at a 90
given value of pressure, the maximum possible equilibrium conver-
sion for ethylene is achieved at the minimum (bubble-point) value
of r. From a practical standpoint, the minimum possible value of 60
pressure for which the equilibrium conversion achieves very high
Vapor
values looks convenient. At 30 atm, those values are achieved for a
water to ethylene feed mole ratio slightly less than 17:1. 30
Chemical equilibrium calculations for the vapor-phase hydra-
tion of ethylene at 200 ◦ C were carried out by solving Eq. (15)
0
with the help of the expression yi = (ni,0 + i εe )/(n0 + εe ). As shown
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
in Fig. 6, for many values of r, it happens that the vapor–liquid
mixed-phase hydration of ethylene to ethanol achieves equilibrium transformed mole fraction of ethylene
conversions much higher than those computed for a vapor-phase
Fig. 7. Reactive phase diagram for the ethylene–water–ethanol ternary system at
reaction that would hypothetically occur at the same conditions of 200 ◦ C. Point C is a reactive critical point, found from a cubic extrapolation of the
temperature, pressure and ethylene to water feed mole ratio. upper parts of the calculated liquid and vapor chemical equilibrium curves.
54 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57

4.5. Reactive phase diagram List of symbols

From the formulation given by Ung and Doherty [109], by choos-

ing ethanol (3) as the reference component, the following set of
transformed composition variables can be defined for ethylene (1) a, b parameters of the PR-EoS
and water (2) in the liquid phase: A = aP/(RT)2 dimensionless attractive parameter of the PR-EoS
B = bP/RT dimensionless repulsive parameter of the PR-EoS
x1 + x3 x2 + x3 Cp0 ideal-gas molar isobaric heat capacity
X1 = , X2 = (20) √ √
1 + x3 1 + x3 C* constant (= [ln( 2 − 1)]/ 2)
Fobj objective function
Analogous expressions can be written for the vapor-phase trans- G
E excess molar Gibbs free energy
formed mole fractions, Y1 and Y2 . Since each pair of transformed I, J integration constants
variables sum to unity, they are not independent, and only one vari- kij binary interaction coefficient
able of each pair is required to plot the reactive phase diagram. From K chemical equilibrium constant
the results given in Table 11, the reactive phase diagram of pressure Ki = yi /xi vapor–liquid equilibrium ratio
versus transformed liquid and vapor-phase ethylene compositions L number of moles in the liquid phase
(P–X1 , Y1 ) at 200 ◦ C was plotted and is shown in Fig. 7. n number of moles
The optimum conditions for the separation of a two-phase nd number of data points
reacting system can be determined from the conceptual design N number of components
of an equilibrium reactive distillation column devised to carry P pressure
out the reaction and separation simultaneously [31,32]. For that Pc critical pressure
design, it is important to know, in the present case, whether P0 standard-state pressure
or not the ethylene–water–ethanol system exhibits a reactive qi molecular surface area parameter
azeotrope at 200 ◦ C. The necessary and sufficient condition for r ethylene to water feed mole ratio
the existence of a reactive azeotrope is the equality X1 = Y1 of ri molecular volume parameter
the transformed mole fractions at a stationary point of both the R universal gas constant
liquid and vapor chemical equilibrium curves [109,110]. In other T absolute temperature
words, these two curves should be tangent at a reactive azeotropic Tc critical temperature
point [111]. In Fig. 7, the liquid and vapor chemical equilibrium Tr reduced temperature
curves join at a pressure of 155 atm, thereby fulfilling the equality V number of moles in the vapor phase
X1 = Y1 = 0.577; however, the curves are not tangent at their join- xi mole fraction in the liquid phase
ing point (point C). In consequence, no reactive azeotrope exists in Xi transformed mole fraction in the liquid phase
the ethylene–water–ethanol system at 200 ◦ C. Point C resembles yi mole fraction in the vapor phase
a critical point for non-reactive systems, which occurs at extrema Yi transformed mole fraction in the vapor phase
of the two-phase envelope on isothermal Pxy plots [112]. In other zi mole fraction in the feed
words, a non-reactive critical point is a tangent point at which the Z PV/RT compressibility factor
Pxy plot is touched by a horizontal line [113]. By analogy, a reac-
tive critical point (like C) is a tangent point at which the PXY plot
Greek symbols
is touched by a horizontal line. For non-reactive systems, the liq-
˛ attractive term of the PR-EoS
uid and vapor phases have the same molar compositions (xi = yi ),

N
i.e. they become indistinguishable, at the mixture critical point. In 0
Cp,rx = 0
i Cp,i standard molar isobaric heat capacity change
contrast, for reacting systems, the liquid and vapor phases have
i=1
the same transformed compositions (Xi = Yi ) at the reactive critical of reaction
point. 0
Gf,i standard molar Gibbs free energy of formation
0
Grx standard molar Gibbs free energy change of reaction
0
Hf,i standard molar enthalpy of formation
5. Conclusions
Hrx 0 standard molar heat of reaction
It was shown here that the PRSV2 + UNIQUAC + WS model uij characteristic interaction energy
is capable of correlating the available experimental VLE data εe equilibrium extent of reaction
at 200 ◦ C for the ethylene–water, ethylene–ethanol, and ϕi pure component fugacity coefficient
(L)
ethanol–water binary systems and predicting the VLE of the ϕ̂i partial fugacity coefficient in the liquid phase
ethylene–water–ethanol ternary system at the same temperature (V )
ϕ̂i partial fugacity coefficient in the vapor phase
and various pressures. The combined CPE of the reacting system ˚i volume fraction
was also predicted and the results were compared with the
i activity coefficient
estimations made in two previous works. It was found that the  parameter of the PRSV2-EoS
reactive phase diagram of the ethylene (1) + water (2) + ethanol (3) i stoichiometric coefficient of component i
ternary system at 200 ◦ C (plotted from the chemical equilibrium 
N
compositions of the liquid and vapor phases), exhibits a reactive = i stoichiometric coefficient of the reaction
critical point at a pressure of 155 atm, where the two phases i=1
have the same transformed molar compositions (X1 = Y1 = 0.577, L liquid fraction
X2 = Y2 = 0.423). Results from the present work can be used to V vapor fraction
study the feasibility of a reactive distillation column in which i surface area fraction
the vapor–liquid mixed-phase catalytic hydration of ethylene to  ij parameter of the UNIQUAC model
ethanol would be carried out simultaneously with the separation ω acentric factor
of the components of the reacting ternary system. 1 equilibrium conversion of ethylene
 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57 55

Subscripts where
i, j, k component indices
m for mixture

N 
N

Q = xi xj (B − A)ij (A.9)
0 initial, in the feed
i=1 j=1

Acknowledgements   
N
1 G
E Ai
D= + xi (A.10)
C∗ RT Bi
Financial support from the School of Chemical Engineering, Uni- i=1
versidad del Valle, Cali, Colombia, through a faculty leave for a
with
sabbatical year (M. Llano-Restrepo), and a teaching assistantship
for graduate students (Y.M. Muñoz-Muñoz) is gratefully acknowl- 1
(B − A)ij = (1 − kij )[(B − A)i + (B − A)j ] (A.11)
edged. 2

G
E 
N

Appendix A. = xi ln
i (A.12)
RT
i=1
The Peng–Robinson (PR) equation of state (EoS) [65] is a two-
The WS mixing rules have only one adjustable parameter, kij , for
parameter EoS, cubic in the molar volume V as follows:
the combining rule given in Eq. (A.11). For chemical and VLE calcu-
RT a lations using equations of state, the partial fugacity coefficient ϕ̂i
P= − (A.1)
V −b V (V + b) + b(V − b) is required. For the combination of the PR-EoS and the WS mixing
rules, an expression for ϕ̂i was obtained in Refs. [68,79] which, for
where
convenience, is given here in terms of dimensionless quantities, as
(RTc )2 follows:
a = 0.457235 ˛ (A.2)  
Pc
1 ∂(nBm )
RTc ln ϕ̂i = − ln(Zm − Bm ) + (Zm − 1)
b = 0.077796 (A.3) Bm ∂ni
Pc     
with Am 1 1 ∂(n2 Am ) 1 ∂(nBm )
+ √ −
2 2Bm Am n ∂ni Bm ∂ni
1/2 2
˛ = [1 + (1 − Tr )] (A.4)  √ 
Zm + (1 − 2)Bm
A better representation of the vapor pressures of pure fluids was ln √ (A.13)
Zm + (1 + 2)Bm
obtained by Stryjek and Vera [66] by correlating  in terms of both
ω and Tr , as follows: where
    
 = 0 , for Tr ≥ 1 (A.5a) 1 ∂(nBm ) 1 1 1 ∂(n2 Q ) Q ∂(nD)
= − 1−
Bm ∂ni Bm 1−D n ∂ni (1 − D)2 ∂ni
=
1/2 1/2
0 + [1 + 2 (3 − Tr )(1 − Tr )](1 + Tr )(0.7 − Tr ), (A.14)

for Tr < 1 (A.5b)      

1 1 ∂(n2 Am ) Bm 1 ∂(nBm ) ∂(nD)
where = D + (A.15)
Am n ∂ni Am Bm ∂ni ∂ni
2 3
0 = 0.378893 + 1.4897153ω − 0.17131848ω + 0.0196544ω
and
(A.6)
1 ∂(n2 Q )  N

The optimum values of 1 , 2 , and 3 were determined for many =2 xj (B − A)ij (A.16)
n ∂ni
fluids in Refs. [66,70,71]. The PR-EoS with the three-adjustable- j=1
parameter form for the attractive term ˛ given by Eqs. (A.4)–(A.6) is
∂(nD) A ln
i
known as PRSV2-EoS. Values of Tc , Pc , ω, 1 , 2 , and 3 for ethylene, = i + (A.17)
∂ni Bi C∗
water, and ethanol are given in Table 5.
Eq. (A.1) can be written as a cubic equation in terms of the com- For the application of the WS mixing rules, an activity coefficient
pressibility factor Z. For a mixture, that equation takes the form model has to be chosen. The UNIQUAC model [67] has been used
3 2 2 3 2 successfully with those mixing rules, for the correlation of VLE
Zm + (Bm − 1)Zm + (Am − 3Bm − 2Bm )Zm + (Bm + Bm − Am Bm ) = 0
data for some binary mixtures of non-polar and polar compounds
(A.7) [75,76,80]. The UNIQUAC model is given by the following expres-
sion [67]:
If Am and Bm are known, Eq. (A.7) can be solved for Zm , by find-
ing the physically admissible real root from the analytical formulas  
of the Cardano-Vieta method [72]. Mixing rules are required to ˚  i
˚ 
N
i i
define Am and Bm for the application of the PR-EoS to mixtures. ln
i = ln + 5qi ln + li − xj lj
xi ˚i xi
The Wong–Sandler (WS) mixing rules [68] have proven to be very j=1

useful for the correlation of VLE experimental data for binary mix- ⎡ ⎤
tures of non-polar and polar compounds at low and high pressures ⎛ ⎞
⎢ ⎥
(see e.g. Refs. [73–78]). In terms of dimensionless parameters, the 
N
⎢  N
j ij ⎥
WS mixing rules are given by the following expressions [68,79]: − qi ln ⎝ j ji ⎠ + qi ⎢
⎢1 − ⎥
⎥ (A.18)
j=1 ⎣ j=1

N
⎦
Q k kj
Bm = , Am = Bm D (A.8)
1−D k=1
56 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57

where [32] H. Hasse, Thermodynamics of reactive separations, in: K. Sundmacher, A.

Kienle (Eds.), Reactive Distillation: Status and Future Directions, Wiley-VCH,
2003, pp. 65–96.
lj = 5(rj − qj ) − (rj − 1) (A.19)
[33] D.S. Tsiklis, E.V. Mushkina, L.I. Shenderei, Inzh. Fiz. Zh. 1 (8) (1958) 3–7.
[34] R.G. Anthony, J.J. McKetta, J. Chem. Eng. Data 12 (1967) 17–20.
xi ri xi qi
˚i = N , i = N (A.20) [35] M. Sanchez, H. Lentz, High Temp.-High Press. 5 (1973) 689–699.
[36] D.S. Tsiklis, A.N. Kofman, Russ. J. Phys. Chem. 35 (1961) 549–551 (original
xr
j=1 j j
xq
j=1 j j Russian version published in Zhur. Fiz. Khim. 35 (1961) 1120–1124).
[37] J. Griswold, J.D. Haney, V.A. Klein, Ind. Eng. Chem. 35 (1943) 701–704.
and [38] F. Barr-David, B.F. Dodge, J. Chem. Eng. Data 4 (1959) 107–121.
  [39] D.S. Tsiklis, A.I. Kulikova, L.I. Shenderei, Khim. Promst. (5) (1960) 401–406.
uij [40] P.A. Gupte, P. Rasmussen, A. Fredenslund, Ind. Eng. Chem. Fundam. 25 (1986)
ij = exp − (A.21) 636–645.
RT [41] J. Schwartzentruber, H. Renon, Ind. Eng. Chem. Res. 28 (1989) 1049–1055.
[42] M.L. Michelsen, Fluid Phase Equilibr. 60 (1990) 47–58.
The characteristic interaction energies uij are the adjustable [43] S. Dahl, M.L. Michelsen, AIChE J. 36 (1990) 1829–1836.
[44] D.S.H. Wong, H. Orbey, S.I. Sandler, Ind. Eng. Chem. Res. 31 (1992) 2033–2039.
parameters of the UNIQUAC model. Values of ri and qi for ethylene,
[45] K. Fischer, J. Gmehling, Fluid Phase Equilibr. 121 (1996) 185–206.
water, and ethanol are given in Table 6. [46] O. Hernández-Garduza, F. García-Sánchez, E. Neau, M. Rogalski, Chem. Eng. J.
79 (2000) 87–101.
[47] D. Constantinescu, A. Klamt, D. Geana, Fluid Phase Equilibr. 231 (2005)
References 231–238.
[48] J.M. Smith, H.C. van Ness, M.M. Abbott, Introduction to Chemical Engineer-
[1] J.E. Logsdon, in: J.I. Kroschwitz, M. Howe-Grant (Eds.), Kirk-Othmer Encyclo- ing Thermodynamics, 7th ed., McGraw-Hill International Edition, 2005, pp.
pedia of Chemical Technology, vol. 9, 4th ed., John Wiley & Sons, New York, 573–575.
1994, pp. 812–860. [49] E. Voutsas, V. Louli, C. Boukouvalas, K. Magoulas, D. Tassios, Fluid Phase Equi-
[2] Chemical Industry News & Intelligence (ICIS.com), Peak practice, ICIS News libr. 241 (2006) 216–228.
(March 16, 1998). Available at http://www.icis.com. [50] Y. Shimoyama, Y. Iwai, S. Takada, Y. Arai, T. Tsuji, T. Hiaki, Fluid Phase Equilibr.
[3] H.A. Wittcoff, B.G. Reuben, J.S. Plotkin, Industrial Organic Chemicals, 2nd ed., 243 (2006) 183–192.
John Wiley & Sons, Hoboken, NJ, 2004, pp. 31, 136. [51] M.-T. Lee, S.-T. Lin, Fluid Phase Equilibr. 254 (2007) 28–34.
[4] Chemical Industry News & Intelligence (ICIS.com), Chemical Profile Ethanol [52] R.H. Ewell, Ind. Eng. Chem. 32 (1940) 147–153.
(January 20, 2003). Available at http://www.icis.com. [53] A.A. Vvedensky, L.F. Feldman, Zh. Obsh. Khim. 15 (1945) 37–41.
[5] Chemical Industry News & Intelligence (ICIS.com), Industrial ethanol market [54] R.R. Wenner, Chem. Eng. Progr. 45 (March) (1949) 194–207.
tightening, ICIS News (June 27, 2005). Available at http://www.icis.com. [55] J. Muller, H.I. Waterman, Gen. Chim. 78 (1957) 173–186.
[6] SADAF (Saudi Petrochemical Co.), In Focus, The SADAF Team: [56] Yu.M. Bakshi, A.I. Gelbshtein, M.I. Temkin, Dokl. Akad. Nauk SSSR 126 (1959)
the Right Chemistry in a Tough Global Arena. Available at 314–317.
http://www.barrowandschuck.com/pdf/12-15%20sadaf%2002.pdf. [57] Yu.M. Bakshi, A.I. Gelbshtein, M.I. Temkin, Dokl. Akad. Nauk SSSR 132 (1960)
[7] M. Bristow, Chemical Industry News & Intelligence (ICIS.com), INEOS restarts 157–159.
synthetic ethanol unit at Grangemouth, ICIS News (June 5, 2009). Available [58] D.S. Tsiklis, A.I. Kulikova, Khim. Promst. (6) (1962) 413–418.
at http://www.icis.com. [59] Thermodynamic Research Center, TRC Thermodynamic
[8] M. Bristow, Chemical Industry News & Intelligence (ICIS.com), Two INEOS’s Tables—Hydrocarbons, Texas A & M University System, College Station, TX,
synthetic ethanol plants running at reduced rate, ICIS News (August 7, 2009). 1985–2002.
Available at http://www.icis.com. [60] D.D. Wagman, W.H. Evans, V.B. Parker, R.H. Schumm, I. Halow, S.M. Bailey,
[9] G. Gilmartin, Chemical Industry News & Intelligence (ICIS.com), Sasol’s Herne K.L. Churney, R.L. Nuttall, J. Phys. Chem. Ref. Data 11 (Suppl. 2) (1982), 2-38,
ethanol plant back onstream, ICIS News (September 15, 2005). Available at 2-95.
http://www.icis.com. [61] C.S. Cope, B.F. Dodge, AIChE J. 5 (1959) 10–16.
[10] C.R. Nelson, M.A.D. Taylor, D.D. Davidson, L.M. Peters (assigned to Shell Devel- [62] J.M. Smith, H.C. van Ness, M.M. Abbott, Introduction to Chemical Engineer-
opment Co.), US Patent 2,579,601 (1951). ing Thermodynamics, 7th ed., McGraw-Hill International Edition, 2005, pp.
[11] C.R. Nelson, M.L. Courter, Chem. Eng. Progr. 50 (October) (1954) 526–531. 511–514.
[12] O.D. Frampton (assigned to National Distillers and Chemical Corp.), US Patent [63] H. Otsuki, F.C. Williams, Chem. Eng. Progr. Symp. Ser. 49 (1953) 55–67.
4,012,452 (1977). [64] O.A. Hougen, K.M. Watson, Chemical Process Principles, Part 2, John Wiley,
[13] R. Devon, M.L. Schwartz, Chem. Eng. 79 (19) (1972) 50–51. New York, 1947, pp. 489, 495, 622.
[14] Shell Development Co., in: T. Ponder (Ed.), Petrochemical Handbook Issue, [65] D.-Y. Peng, D.B. Robinson, Ind. Eng. Chem. Fundam. 15 (1976) 59–64.
Hydrocarbon Process. 52 (November) (1973) 121. [66] R. Stryjek, J.H. Vera, Can. J. Chem. Eng. 64 (1986) 820–826.
[15] Union Carbide Corp., in: T. Ponder (Ed.), Petrochemical Handbook Issue, [67] D.S. Abrams, J.M. Prausnitz, AIChE J. 21 (1975) 116–128.
Hydrocarbon Process. 52 (November) (1973) 122. [68] D.S.H. Wong, S.I. Sandler, AIChE J. 38 (1992) 671–680.
[16] Veba-Chemie AG, in: T. Ponder (Ed.), Petrochemical Handbook Issue, Hydro- [69] K. Eguchi, T. Tokiai, Y. Kimura, H. Arai, Chem. Lett. (Japan) 15 (1986) 567–570.
carbon Process 54 (November) (1975) 135. [70] R. Stryjek, J.H. Vera, Can. J. Chem. Eng. 64 (1986) 323–333.
[17] A.E. Sommer, R. Bücker, in: J.J. McKetta, W.A. Cunningham (Eds.), Encyclopedia [71] P. Proust, J.H. Vera, Can. J. Chem. Eng. 67 (1989) 170–173.
of Chemical Processing and Design, vol. 19, Marcel Dekker, New York, 1983, [72] M.R. Spiegel, J. Liu, Mathematical Handbook of Formulas and Tables, 2nd ed.,
pp. 452–455. McGraw-Hill, 1999, p. 10.
[18] T.C. Carle, D.M. Stewart, Chem. Ind. (London) 19 (1962) 830–839. [73] G.-S. Shyu, N.S.M. Hanif, K.R. Hall, P.T. Eubank, Fluid Phase Equilibr. 130 (1997)
[19] A.F. Millidge, Catalytic hydration, in: S.A. Miller (Ed.), Ethylene and its Indus- 73–85.
trial Derivatives, Ernest Benn Ltd., London, 1969, pp. 709–731. [74] P. Ghosh, T. Taraphdar, Chem. Eng. J. 70 (1998) 15–24.
[20] R.C. Thomson, P.W. Reynolds (assigned to Imperial Chemical Industries Ltd.), [75] T. Ohta, M. Ishio, T. Yamada, Fluid Phase Equilibr. 153 (1998) 105–111.
British Patent 665,214 (1952). [76] A.M. Scurto, C.M. Lubbers, G. Xu, J.F. Brennecke, Fluid Phase Equilibr. 190
[21] R.C. Thomson, R.K. Greenhalgh (assigned to Imperial Chemical Industries Ltd.), (2001) 135–147.
British Patent 691,360 (1953). [77] J.A. López, V.M. Trejos, C.A. Cardona, Fluid Phase Equilibr. 248 (2006) 147–
[22] P.W. Reynolds, L.R. Pittwell (assigned to Imperial Chemical Industries Ltd.), 157.
US Patent 2,755,309 (1956). [78] J.A. López, V.M. Trejos, C.A. Cardona, Fluid Phase Equilibr. 275 (2009) 1–7.
[23] A.W.C. Taylor (assigned to Imperial Chemical Industries Ltd.), US Patent [79] H. Orbey, S.I. Sandler, Modeling Vapor–liquid Equilibria: Cubic Equations of
2,815,391 (1957). State and their Mixing Rules, Cambridge University Press, 1998, pp. 106–107.
[24] P. Bazzarin (assigned to Montecatini), US Patent 3,164,641 (1965). [80] H. Huang, S.I. Sandler, Ind. Eng. Chem. Res. 32 (1993) 1498–1503.
[25] R. Kolodziej, S. Dutt, Hydrocarbon Process. 80 (October) (2001) 103–110. [81] J.M. Prausnitz, T.F. Anderson, E.A. Grens, C.A. Eckert, R. Hsieh, J.P. O’Connell,
[26] Chemical Industry News & Intelligence (ICIS.com), Equistar heightens product Computer Calculations for Multicomponent Vapor–liquid and Liquid–liquid
purity, ICIS News (April 13, 1998). Available at http://www.icis.com. Equilibria, Prentice Hall, New Jersey, 1980 (Appendix C-1).
[27] J. Kamalick, Chemical Industry News & Intelligence (ICIS.com), Kvaerner to [82] J.R. Elliot, C.T. Lira, Introductory Chemical Engineering Thermodynamics,
market Lyondell alcohol tech, ICIS News (November 10, 1999). Available at Prentice Hall, New Jersey, 1999, pp. 336, 617.
http://www.icis.com. [83] S.I. Sandler, Chemical, Biochemical, and Engineering Thermodynamics, 4th
[28] A.I. Stankiewicz, J.A. Moulijn, Chem. Eng. Progr. 96 (January) (2000) 22–34. ed., John Wiley & Sons, New Jersey, 2006, pp. 562, 563.
[29] A. Stankiewicz, J.A. Moulijn, Ind. Eng. Chem. Res. 41 (2002) 1920–1924. [84] H.H. Rachford, J.D. Rice, Petrol. Trans. AIME 195 (1952) 327–328.
[30] M.M. Sharma, S.M. Mahajani, Industrial applications of reactive distillation, [85] M.L. Michelsen, Fluid Phase Equilibr. 9 (1982) 1–19.
in: K. Sundmacher, A. Kienle (Eds.), Reactive Distillation: Status and Future [86] M.L. Michelsen, Comp. Chem. Eng. 17 (1993) 431–439.
Directions, Wiley-VCH, 2003, p. 26. [87] J.A. Nelder, R. Mead, Comp. J. 7 (1965) 308–313.
[31] M.F. Doherty, M.F. Malone, Conceptual Design of Distillation Systems, [88] G.N. Vanderplaats, Numerical Optimization Techniques for Engineering
McGraw-Hill International Edition, 2001, pp. 427–505. Design with Applications, McGraw-Hill, New York, 1984, pp. 80–83.
 M. Llano-Restrepo, Y.M. Muñoz-Muñoz / Fluid Phase Equilibria 307 (2011) 45–57 57

[89] H. Huang, S.I. Sandler, H. Orbey, Fluid Phase Equilibr. 96 (1994) 143–153. [111] D. Barbosa, M.F. Doherty, Proc. R. Soc. Lond. A 413 (1987) 459–464.
[90] J.M. Smith, H.C. van Ness, M.M. Abbott, Introduction to Chemical Engineer- [112] J.P. O’Connell, J.M. Haile, Thermodynamics: Fundamentals for Applications,
ing Thermodynamics, 7th ed., McGraw-Hill International Edition, 2005, p. Cambridge University Press, 2005, pp. 377–379, 388.
566. [113] J.M. Smith, H.C. van Ness, M.M. Abbott, Introduction to Chemical Engineer-
[91] B.G. Kyle, Chemical and Process Thermodynamics, 3rd ed., Prentice Hall, New ing Thermodynamics, 7th ed., McGraw-Hill International Edition, 2005, pp.
Jersey, 1999, p. 297. 342–343.
[92] J.M. Smith, H.C. van Ness, M.M. Abbott, Introduction to Chemical Engineer-
ing Thermodynamics, 7th ed., McGraw-Hill International Edition, 2005, pp.
Mario Llano-Restrepo is a professor of Chemical Engi-
140–141, 490–492, 498, 684, 686–687.
neering at Universidad del Valle in Cali, Colombia. He
[93] J. Carrero-Mantilla, M. Llano-Restrepo, Fluid Phase Equilibr. 219 (2004)
received his Ph.D. degree (in Chemical Engineering) from
181–193.
Rice University in 1994. His main research interests are
[94] H.M. Spencer, Ind. Eng. Chem. 40 (1948) 2152–2154.
modeling of phase and chemical equilibria, molecular
[95] L.B. Pankratz, Thermodynamic Properties of Elements and Oxides, U.S. Bureau
simulation, and modeling and simulation of separation
of Mines Bulletin 672, US GPO, Washington DC, 1982, p. 182.
processes. In 2009, he earned a Distinguished Professor
[96] F.J. Sanders, B.F. Dodge, Ind. Eng. Chem. 26 (1934) 208–214.
recognition from Universidad del Valle for service and
[97] H.M. Stanley, J.E. Youell, J.B. Dymock, J. Soc. Chem. Ind. Trans. Commun. 53
excellence in teaching during the past 15 years. Among
(1934) 205–208.
the courses he has taught are chemical thermodynamics,
[98] E.R. Gilliland, R.C. Gunness, V.O. Bowles, Ind. Eng. Chem. 28 (1936) 370–372.
statistical thermodynamics, molecular simulation, chemi-
[99] R.H. Bliss, B.F. Dodge, Ind. Eng. Chem. 29 (1937) 19–25.
cal reaction engineering, separation process modeling and
[100] B.F. Dodge, Ind. Eng. Chem. 29 (1937) 845–846.
simulation, advanced mathematics, advanced transport
[101] M.P. Applebey, J.V.S. Glass, G.F. Horsley, J. Soc. Chem. Ind. Trans. Commun. 56
phenomena, and reactor analysis and design.
(1937) 279–281.
[102] J.G. Aston, Ind. Eng. Chem. 34 (1942) 514–521.
[103] R.V. Sanderson, H.H.-Y. Chien, Ind. Eng. Chem. Process Des. Dev. 12 (1973) Y. Mauricio Muñoz-Muñoz is currently a Ph.D. student
81–85. at the School of Chemical Engineering, Universidad del
[104] W.-D. Xiao, K.-H. Zhu, W.-K. Yuan, H.H.-Y. Chien, AIChE J. 35 (1989) Valle in Cali, Colombia. As a teaching assistant for the
1813–1820. chemical engineering program of study, he has taught the
[105] J.P. O’Connell, J.M. Haile, Thermodynamics: Fundamentals for Applications, mass transfer course for the past two years. He received
Cambridge University Press, 2005, p. 513. his undergraduate degree (B.S.) in Chemical Engineering
[106] R.P. Stateva, W.A. Wakeham, Ind. Eng. Chem. Res. 36 (1997) 5474–5482. from Universidad Nacional de Colombia, at the Manizales
[107] S.M. Walas, Phase Equilibria in Chemical Engineering, Butterworth- campus, in 2007. His main research interests are molecu-
Heinemann, Stoneham, MA, 1985, pp. 494–495. lar simulation, thermodynamics of irreversible processes,
[108] J.M. Smith, H.C. van Ness, M.M. Abbott, Introduction to Chemical Engineer- and intensification of separation processes.
ing Thermodynamics, 7th ed., McGraw-Hill International Edition, 2005, pp.
514–515.
[109] S. Ung, M.F. Doherty, Chem. Eng. Sci. 50 (1995) 23–48.
[110] S. Ung, M.F. Doherty, AIChE J. 41 (1995) 2383–2392.