Brazilian Journal

of Chemical ISSN 0104-6632

Printed in Brazil
Engineering www.abeq.org.br/bjche

Vol. 32, No. 04, pp. 957 - 966, October - December, 2015
dx.doi.org/10.1590/0104-6632.20150324s20140023

THERMODYNAMIC TOPOLOGICAL ANALYSIS

OF EXTRACTIVE DISTILLATION OF MAXIMUM
BOILING AZEOTROPES
W. F. Shen1,2, H. Benyounes3* and J. Song4
1
Université de Toulouse, INP, UPS, LGC (Laboratoire de Génie Chimique),
4 allée Emile Monso, F-31432 Toulouse Cedex 04, France.
2
CNRS, LGC (Laboratoire de Génie Chimique), F-31432 Toulouse Cedex 04, France.
3
U.S.T. Oran, Laboratoire de Chimie Physique des Matériaux, Catalyse et Environnement, Oran, Algérie.
E-mail: bhassba@yahoo.fr
4
National Engineering Laboratory for Hydrometallurgical Cleaner Production Technology,
Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100190, China.

(Submitted: August 23, 2014 ; Revised: October 14, 2014 ; Accepted: November 25, 2014)

Abstract - This paper provides a feasibility study of azeotropic mixture separation based on a topological
analysis combining thermodynamic knowledge of residue curve maps, univolatility and unidistribution
curves, and extractive profiles. Thermodynamic topological features related to process operations for typical
ternary diagram classes 1.0-2 are, for the first time, discussed. Separating acetone/chloroform is presented as
an illustrative example; different entrainers are investigated: several heavy ones, a light one, and water,
covering the Serafimov classes 1.0-2, 1.0-1a and 3.1-4, respectively. The general feasibility criterion that was
previously established for ternary mixtures including only one azeotrope (1.0-1a or 1.0-2) is now, for the first
time, extended to that including three azeotropes (class 3.1–4).
Keywords: Azeotropic mixtures; Topological analysis; Residue curve map; Extractive distillation; Maximum
boiling azeotrope.

INTRODUCTION are usually calculated from experimental data or

from thermodynamic activity coefficient models
In most separation systems, the predominant non- regressed on experimental data. Azeotropes occur in
ideality occurs in the liquid phase due to molecular a number of nonideal systems. An azeotrope exists
interactions. When chemically dissimilar compo- when the liquid and vapor compositions are the same
nents are mixed together (e.g., oil molecules and (xi=yi) at a given azeotrope temperature. Figure 1
water molecules), there exists repulsion or attraction and 2sketch typical graphical phase representations
between dissimilar molecules. If the molecules repel of the vapor-liquid equilibrium (VLE).The left part
each other, they exert a higher partial pressure than if of Figures 1 and 2 shows a combined graph of the
the mixtures were ideal. In this case, the activity bubble and dew temperatures, pressure, and the
coefficients are greater than unity (called a “positive vapor-liquid equilibrium phase mapping, which give
deviation” from Raoult‘s law). If the molecules at- a complete representation of the VLE. In addition,
tract each other, they exert a lower partial pressure the right parts these figures give the equilibrium
than in an ideal mixture. Activity coefficients are less phase mapping y vs. x alone. Each of these diagrams
than unity (negative deviations). Activity coefficients uniquely characterizes the type of corresponding

*To whom correspondence should be addressed

958 W. F. Shen, H. Benyounes and J. Song

  Pure A

vapor P
Tazeo  KA>1  
P 0A
 
TB
liquid liquid A rich
  in
T yA
P0B B rich in vapor phrase
vapor
Pazeo phrase
TA vapor Azeotrope  
KA =1
xazeo xazeo
KA<1
xA and yA xA and yA Pure B xA 

Figure 1: Typical homogeneous mixtures with maximum boiling azeotrope.

  
Pure A
P P
Liquid Liquid yA
Pazeo Pazeo Azeotrope
B rich in
vapor
P0A P0A
phase

P0B P0B
Vapor Vapor
A rich in
xazeo xazeo vapor phase

Pure B xA
xA and yA xA and yA

Figure 2: Typical homogeneous mixtures with minimum boiling azeotrope.

mixture. Negative deviations (attraction) can give a 0 or 1. These classes are further divided into types
higher temperature boiling mixture than the boiling and subtypes denoted by a number and a letter. As a
point of the heavier component, called a maximum- result of this detailed analysis, 4 more feasible topo-
boiling azeotrope (Figure 1). Positive deviations (re- logical structures, not found by Gurikov, were re-
pulsion) can give a lower temperature boiling mix- vealed. Thus, Serafimov’s classification includes 26
ture than the boiling point of the light component, classes of feasible topological structures of VLE
called a minimum boiling azeotrope (Figure 2). diagrams for ternary mixtures. Both the classifica-
The study of the thermodynamic classification of tions of Gurikov and Serafimov consider topological
liquid-vapor phase equilibrium diagrams for ternary structures and thus do not distinguish between an-
mixtures and their topological interpretation has a tipodal structures since they have the same topology.
long history. Considering a ternary diagram A-B-E Thus, the above classifications include ternary mix-
formed by a binary mixture A-B with the addition of tures with opposite signs of the singular points and in
an entrainer E, the classification of azeotropic mix- an opposite direction of the residue curves (antipodal
tures into 113 classes was first proposed by Matsu- diagrams). Serafimov’s classification is presented
yama et al. (1977). As explained by Hilmen (2000), graphically in Figure 3. The transition from one anti-
Serafimov (1996) extended the work of Gurikov pode to the other (e.g., changing from minimum-to
(1958) and used the total number of binary azeotropes maximum-boiling azeotropes) can be made by
M and the number of ternary azeotropes T as classifi- simply changing the signs of the nodes and inverting
cation parameters. Serafimov’s classification denotes the direction of the arrows. The correspondence be-
a structural class by the symbol “M.T.” where M can tween Matsuyama and Serafimov’s classification is
take the values 0, 1, 2 or 3, and T can take the values detailed in Kiva et al. (2003).

Brazilian Journal of Chemical Engineering

 Thermodynamic Topological Analysis of Extractive Distillation of Maximum Boiling Azeotropes 959

2013b; Laroche et al., 1992). Based on the calcu-

lation of the still path and possible composition pro-
files of the column sections, Lang et al. (2000a, b)
extended the feasibility of extractive distillation in a
batch rectifier for the investigation of the separation
of maximum azeotropes, Rodriguez-Donis et al.
(2009a, 2012a,b) studied the maximum azeotropic
mixture separation. However, extractive distillation
studies have always considered a heavy entrainer to
split a minimum boiling azeotrope which belongs to
class 1.0-1a (Knapp and Doherty, 1994a; Luyben,
2008a, 2008b; Brüggemann and Marquardt, 2004).
Düssel and Stichlmair (1995) proposed a hybrid
method by replacing one distillation step into a de-
cantation or absorption, which is less energy inten-
sive for the separation of minimum azeotropes with a
heavy entrainer. These study focus on maximum
azeotropic mixture separation and provide a useful
guide for the separation of azeotropic mixtures by
batch extractive distillation.

SIMPLE THERMODYNAMIC INSIGHT

TOOLS

Residue Curve Map

Figure 3: Azeotropic ternary mixture: Serafimov’s A residue curve map (RCM) is a collection of the
26 topological classes and Reshetov’s statistics (o) liquid residue curves in a simple one-stage batch
unstable node, (∆) saddle, (●) stable. Reproduced from distillation originating from different initial composi-
Hilmen et al. (2003), with permission from Elsevier. tions. The RCM technique is considered to be a pow-
erful tool for flow-sheet development and the pre-
Ternary systems are studied in this research on liminary design of conventional multi-component
the basis of Serafimov’s classification, including 26 separation processes and has been extensively stud-
classes of feasible topological structures of VLE dia- ied since 1900. Using the theory of differential equa-
grams for ternary mixtures (Serafimov, 1996). The tions, the studies of the topological properties of the
entrainer E is conventionally defined by its boiling residue curve map (RCM) are summarized in two
temperature with respect to the binary mixture A-B recent articles (Kiva et al., 2003; Hilmen et al., 2003).
to separate. A heavy entrainer E has a boiling tem- The simple RCM was modeled by the set of dif-
perature higher than A and B; an intermediate en- ferential equations.
trainer E has a boiling temperature between the A
and B; a light entrainer E has a boiling temperature dxi
lower than A and B. In industry, the extractive distil-  xi  yi (1)
dh
lation entrainer is usually chosen as heavy (high boil-
ing) component mixtures (Luyben and Chien, 2010; where h is a dimensionless time describing the rela-
Lang et al., 1994; Rodriguez-Donis, 2009; Shen 2012; tive loss of the liquid in the still-pot.xi is the mole
Shen et al., 2013a; Benyounes et al., 2014; Benyahia fraction of species i in the liquid phase, and yi is the
et al., 2014). Theoretically, any candidate entrainer mole fraction of species i in the vapor phase. The yi
satisfying the feasibility and optimal criteria can be values are related with the xi values using equilib-
used, no matter whether it is a heavy, light, or inter- rium constants Ki.
mediate entrainer. Literature studies on intermediate The singular points of the differential equation
entrainers or light entrainers validate this assumption are checked by computing the associated eigen-
(Lelkes et al., 2002; Lang et al., 1999; Rodriguez- values. Within anon-reactive residue curve map, a
Donis et al., 2012; Shen, 2012; Shen and Gerbaud, singular point can be a stable or an unstable node or

Brazilian Journal of Chemical Engineering Vol. 32, No. 04, pp. 957 - 966, October - December, 2015
960 W. F. Shen, H. Benyounes and J. Song

a saddle, depending on the sign of the eigenvalues the behavior of these functions for binary mixtures.
related to the residue curve equation. For non-reac- The composition dependencies of the distribution
tive mixtures, there are 3 stabilities: unstable node, coefficients are qualitative and quantitative charac-
stable node, and saddle point. The residue curves teristics of the VLE for the given mixture. In a simi-
move away from the unstable node to stable node lar way to the distribution coefficient, the relative
with increasing temperatures. Some residue curves volatility features can be represented by isovolatility
move away from a saddle point with decreasing tem- lines and the system of univolatility lines when αij=1
peratures and others with increasing temperatures. proposed. It is evident that the point of a binary aze-
otrope gives rise to aunivolatility line and then the
Unidistribution and Relative Volatility point of a ternary azeotrope gives rise to the three
univolatility lines. These features are represented in
The distribution coefficient and relative volatility Figure 4 for the most probable classes. The main aim
are well-known characteristics of the vapor–liquid of their work (Kiva et al., 2003) was to consider
equilibrium. The distribution coefficient Ki is defined feasible structures of the residue curve maps in more
by: detail, and in fact this study helped to popularize
more refined classification of the ternary diagrams.
yi The diagrams of unidistribution lines were used as a
Ki  (0) main tool for analysis of tangential azeotropes.
xi

Ki characterizes the distribution of component i

between the vapor and liquid phases in equilibrium.
Ki=1 defines the unidistribution curve. The vapor is
enriched with component i if Ki>1, and is impover-
ished with component i if Ki<1 compared to the liquid.
The higher Ki, the greater the driving force (yi-xi) and
the easier the distillation will be. The ratio of the
distribution coefficient of components i and j gives
the relative volatility. The relative volatility is a very
convenient measure of the ease or difficulty of sepa-
ration in distillation. The volatility of component j
relative to component i is defined as:

y i / xi
 ij  (3)
yj / xj

The relative volatility characterizes the ability of

component i to transfer (evaporate) into the vapor
phase compared to the ability of component j. Com-
ponent i is more volatile than component j if  ij >1,
Figure 4: Unidistribution and univolatility line dia-
and less volatile if  ij <1. For ideal and nearly ideal grams for the most probable classes of ternary mix-
mixtures, the relative volatilities for all pair of com- tures according to Reshetov’s statistics. Reproduced
ponents are nearly constant in the whole composition from Kiva et al. (2003), with permission from Elsevier .
space. The situation is different for non-ideal and in
particular azeotropic mixtures, where the composi- Recently, Rodriguez-Donis et al., (2009a, 2009b,
tion dependence can be complex. 2010, 2012a, 2012b) studied how univolatility lines
Unidistribution and univolatility line diagrams split the composition triangle into regions of a cer-
can be used to sketch VLE diagrams and represent tain order of volatility of components and defined a
topological features of the simple phase transfor- general feasibility criterion for extractive distillation
mation trajectories. The qualitative characteristics of under an infinite reflux ratio: “Homogeneous extrac-
the distribution coefficient and relative volatility tive distillation of a A-B mixture with entrainer E
functions are typical approaches for thermodynamic feeding is feasible if there exists a residue curve con-
topological analysis. Kiva et al., (2003) considered necting E to A or B following a decreasing (a) or

Brazilian Journal of Chemical Engineering

 Thermodynamic Topological Analysis of Extractive Distillation of Maximum Boiling Azeotropes 961

increasing (b) temperature direction inside the region 5(b) show the typical feature for a residue curve map
where A or B are the most volatile (a) or the heaviest (Figure 5(a)) and an extractive map (Figure 5(b)).
(b) component of the mixture”. In this work, we con- The column has a single rectifying section, and the
sider unidistribution and univolatility line diagrams composition profile of the rectifying section in a tray
for the purpose of sketching the volatility order re- column follows strictly a residue curve assuming the
gion and thus of assessing the feasible structures that constant molar overflow hypothesis and infinite
will obtain possible products and offer information number of trays (Jobson et al., 1995; Krolikowski,
of possible limitations of entrainer feed. 2002). The rcm analysis states that both original
components A and B are unstable nodes; the en-
APPLICATION OF TOPOLOGIC ANALYSIS trainer (E) is the stable node, while the maximum
FOR SEPARATING ACETONE/CHLOROFORM boiling azeotrope between AB is a saddle point. The
rcm stable separatrix, which is also called the basic
The separation of the maximum-boiling azeotrope distillation region boundary, links the azeotrope to E.
acetone/chloroform using 7 different candidate sol- Separation of components A and B is theoretically
vents was employed as case study. The azeotropic be- impossible by conventional azeotropic distillation
haviors in multicomponent mixtures with solvents adding E initially into the still, because components
are shown in Table 1. Separation of the azeotropic A and B are located in different distillation regions,
mixture A-B with heavy entrainers DMSO (E1), chlo- separated by the rcm stable separatrix. Under infinite
robenzene (E2), EG (E3), o-xylene (E4) or benzene (E5) R and FE/V~0+ (Figure 5(b)), the maximum boiling
has same type of singular properties, as they exhibits azeotrope azeoAB is a saddle Sextr and A and B are
the same 1.0-2 classification, while using water (E6) stable extractive nodes (SNA,extr and SNB,extr, respec-
introduces two more azeotropes, which makes the tively), whereas E is an unstable extractive node
separation more complicated. The light entrainer (UNextr). There will always be an unstable extractive
dichloromethane (E7) has another type of singular separatrix between UNextr (vertex E) and Sextr (Tmax
character. The separation of the maximum-boiling azeotrope AB). All the general features of the to-
mixture acetone/chloroform with the proposed sol- pology of the extractive composition profile map and
vents exhibit class 1.0-1a, 1.0-2, and 3.1-4. The ther- its difference relative to class 1.0-1a are now dis-
modynamic insights published in previous work cussed as follows, depending on the intersection of
(Shen et al., 2013a; Benyounes et al., 2014; Benyahia the αAB=1 curve with the triangle edges. Figure 5(c)
et al., 2014; Lelkes et al., 2002; Lang et al., 1999; shows the extractive composition profile maps for a
Rodriguez-Donis et al., 2012; Shen et al., Gerbaud, higher value of FE/V but lower than (FE/V)max while
2013b) and validated for the 1.0-1a and 1.0-2 ternary R is infinite. Sextr moves inside the ternary composi-
mixture class are applied here, and the general feasi- tion space, precisely along the univolatility line αAB=1.
bility criterion previously established for ternary Furthermore, the stable extractive nodes SNA,extr and
mixtures including only one azeotrope (1.0-1a or SNB,extr move toward E over the binary edges A-E and
1.0-2) is now, for the first time, extended to that in- B-E, respectively. Therefore, there exist a stable
cluding three azeotropes (class 3.1–4). extractive separatrix SNB,extr-Sextr-SNA,extr and an
unstable extractive separatrix UNextr-Sextr-UNextr’.
Topological Features Related to Process Operation Logically, under finite reflux ratio, the unstable
extractive separatrix UNextr-Sextr-UN’ will move
The general feasibility criterion holds for infinite toward the selected distillate product (A or B),
reflux ratio operations. In this section we take a class reducing the size of their respective feasible regions.
1.0-2 diagram to display the qualitative topological Further increases in FE/V allow the fusion of Sextr and
features of the ternary system related to process opera- SNA,extr. All extractive composition profiles then reach
tion. Under infinite R and FE/V=0+, Figures. 5(a) and the unstable node SNB,extr (Figure 5(d)).

Table 1: Types of single points of extractive distillation entrainers.

Azeotropes introduced by E E A B AZEO_AB

A-E B-E A-B-E
Heavy entrainers (E1- E5) none none none Stable Unstable Unstable Saddle
Light entrainer (E7) none none none Unstable Saddle Saddle Stable
Water (E6) Unstable Unstable Saddle Stable Saddle Saddle Stable

Brazilian Journal of Chemical Engineering Vol. 32, No. 04, pp. 957 - 966, October - December, 2015
962 W. F. Shen, H. Benyounes and J. Song

1.0-2 Residue curve map Extractive profile maps

(a) B [UNrcm] (b) B [SNext,B]
R R
Residue FE/V = 0 FE/V  0+
curve
Distillation
Boundary
I Tmax azeoAB Tmax azeoAB
[Srcm] [Sext]

I AB = 1

E (heavy) A E xP A
[SNrcm] [UNrcm] [UNext] [SNext,A]

(c) B (d) B
R R
FE/V < (FE/V)max,R FE/V > (FE/V)max,R

[SNext,B]
Tmax azeoAB Tmax azeoAB
[SNext,B]

[UNext] [Sext] [UNext]

xP [Sext]
E [SNext,A] A E residue curve A
passing near A

Product A Product B
(e) B (f) B
R finite R finite
FE/V  0+ [SNB,ext] FE/V  0+

[Sext]
Tmax azeoAB Tmax azeoAB
[Sext]
[UNext] [UNext]

xP [SNA,ext] A xP
E E A

Product A Product B
(g) B (h) B
R finite R finite
FE/V < (FE/V)max FE/V < (FE/V)max

Tmax azeoAB Tmax azeoAB

[Sext] [SNB,ext]

[UNext] [UNext]

E [SNA,ext] xP A E [Sext] xP A

Extractive Stable extractive Unstable extractive

profile separatrix separatrix

Figure 5: Topological features related to the extractive distillation process opera-

tion of class 1.0-2 when using a heavy entrainer.

Brazilian Journal of Chemical Engineering

 Thermodynamic Topological Analysis of Extractive Distillation of Maximum Boiling Azeotropes 963

The above observation shows the significance of A and B are rcm unstable nodes. The size of the
the univolatility line in the synthesis of the homo- volatility order regions BAE (and BEA eventually)
geneous extractive distillation process, because it and ABE (and AEB eventually) depends on the αAB
sets limiting values of (FE/V)max. Here, the αAB=1 =1 univolatility curve location. The other univolatility
line sets a maximum value (FE/V)max,B,R∝ to recover curve (αBE =1 in Figure 6(a)) may exist, but does not
component A. Under finite reflux ratio, extractive affect the process feasibility and the product
profiles are dependent on the distillate composition. targeting. Both A and B are connected by a residue
Therefore, rectifying and extractive composition curve of decreasing temperature to E, which nears
profile maps must be computed for both possible the triangle edge in the ternary diagram. Therefore,
distillates xDA and xDB for different FE/V and R con- they can both be distillate products, depending on the
ditions (Figures 5(e) and 5(g) for product A and Figure location of the global feed composition, either in
5(f) and 5(h) for product B). For Figure 5(e), the BAE (B product) or in ABE and AEB (A product).
ternary saddle Sextr moves toward vertex B while for The extractive profile map analysis enables identi-
Figure 5(g) Sextr moves toward the B-E edge inside fication of feasible and unfeasible regions for the
the composition triangle and the stable node SNB, extr composition in the extractive section of the column.
is located outside the composition triangle. The re- Those regions are bounded by extractive stable and
flux ratio allows one to set the extractive separatrix unstable separatrices crossing at saddle extractive
on the left of the rectifying separatrix, and an optimal singular points (Knapp and Doherty, 1994).The rcm
reflux ratio exists for a given FE/V. The occurrence singular points become the singular points of the
of an unstable extractive separatrix prevents com- extractive profile map with an opposite stability for
plete recovery of the distillate, because an unfeasible the heavy entrainer case. Then for the 1.0-2 class,
region of growing size arises as R decreases. In con- increasing the entrainer flowrate, the extractive pro-
trast, for Figures 5(f) and 5(h), the ternary saddle Sextr file singular points SNextr,A and SNextr,B originating
moves toward the A-E edge inside the composition from A and B move towards the entrainer vertex. In
triangle and the stable node SNA,extr is located outside Figure 6(a), SNextr,A can go up to E and there is no
the composition triangle. limiting flowrate. But SNextr,B disappears at point xP
when it merges with the saddle extractive Sext origi-
Topological Feasibility Criterion nating at the Tmax azeAB. So, there is a maximum
flowrate ratio FE/Vmax to obtain A as product by ex-
Figure 6 displays the general thermodynamic prin- tractive distillation. That behavior is directly related
ciples of the 1.0-2 class corresponding to extractive to the volatility order regions, which explains how
distillation of a maximum-boiling azeotropic mixture the general criterion is established. The above crite-
A-B with heavy entrainer benzene.While using rion indicates that A can be distillated without any
heavy entrainer DMSO, chlorobenzene, EG or xy- limit for the entrainer feed ratio, whereas there exists
lene, the ternary systems bear the same features as a maximum entrainer feed ratio to get B. Depending
benzene. The αAB=1 reaches the binary side B-E when on the overall feed composition, either A or B can be
using these heavy entrainers. Figure 6(a) shows that distilled: from xF1 B is the distillate product; from xF2
a rectifying stable separatrix divides the composition below the extractive separatrix, A is the distillate
space into two distillation regions. Both components product.
  (a) (b)
Chlorform (B) 61.2°C Chlorform (B) 61.2°C FE/V =0.1
Rectifying [SNextr,B] xDB R=60
profile for B XYZ Volatility order Extractive profile
[SNextr,B] Tmax azeoAB
65.1°C (X possible distillate) XF1 Tmax azeoAB
range
BAE 65.1°C
rcm stable
separatrix II [Sextr]
αAB = 1
αAB = 1
xP IV III
xP
ABE Rectifying
αBE = 1 XF2 [SNextr,A]
profile for A
xDA
AEB αBE = 1

Benzene (E) 81.1°C [SNextr,A] Acetone (A) 56.3°C Benzene (E) 81.1°C Acetone (A) 56.3°C
range

Rectifying Separatrix Rectifying Separatrix

  Extractive Separatrix  
Figure 6: Thermodynamic feasibility criterion for 1.0–2 case b with heavy
entrainer benzene (DMSO; chlorobenzene; EG or o-xylene).
Brazilian Journal of Chemical Engineering Vol. 32, No. 04, pp. 957 - 966, October - December, 2015
964 W. F. Shen, H. Benyounes and J. Song

Figure 7(a) summarizes the topological features due curve of increasing temperature from E to A.
and the products achievable for the class 1.0–1a Figure 7c presents the extractive profiles under the
(maximum boiling azeotrope with a light entrainer). condition of infinite reflux and FE/V = 0.5; all the
The point of entrainer dichloromethane is the residue extractive profiles start from the unstable node point,
curve unstable node UNrcm (open circle); apex A and depend on the overall feed composition, and either
B are residue curve saddle points Srcm (open triangle product can be distillated correspondingly.
down), while the maximum boiling azeotrope is a Figure 7(b) displays the thermodynamic proper-
stable node SNrcm (filled circle). By using azeotropic ties for the ternary mixtures of acetone – chloroform
distillation, it is impossible to recover A or B, but – water where the entrainer water forms two addi-
rather the unstable node maxT azeotrope at the col- tional binary minimum-boiling azeotropes. Accord-
umn bottom in a batch stripper or in the continuous ing to Serafimov’s classification they correspond to
column bottom. By using extractive distillation, ei- the diagrams class 3.1-4. Depending on the stability
ther A or B can be removed as product depending on of singular points, there is only one or several possi-
the univolatility line AB=1 location. The univolatil- bilities for the location of the univolatility lines al-
ity line starts at the maximum boiling azeotrope, ways associated with the position of azeotropes.
intersects one triangle side at xp,A and divides the Univolatility lines intercept each other at the ternary
composition graph into two volatility order regions, azeotrope, and part of univolatility lines overlap with
EBA and EAB. The xP,A location decides that A can the residue curve map boundary. The univolatility
be recovered from the bottom as xP,A lies on the A-E lines then define volatility order regions like AEB: A
side (Figure 7(a)). A is the most difficult volatile in is more volatile than E, and E is more volatile than
the region EBA where it is connected to E by a resi- B. From one side of a univolatility line αAB to the

(a) Acetone (A)

(b) Chlorform (B) 61.2°C
(329.2K) [Srcm] FE/LT< (FE/ LT) min
   
XWA S XYZ Volatility order
    (X possible distillate)
Residue curves
Tmin azeoAB
ZYX Volatility order   55.9°C EBA
(X possible bottom) Tmax azeoAB
Tmax azeoAB 64.6°C
EAB αBE = 1
337.3 K αAB = 1
[SNrcm] αAB = 1 BEA
xp,A αAE = 1
[SNextr,A] 60.5°C Tmin azeoAB
EBA range EAB 56.1°C
AEB
XDA

Chloroform (B) Dichloromethane (E) xp

(334.2K) [Srcm] (312.8K) [UNrcm] Water(E)100°C [SNextr,A] Acetone (A) 56.3°C
range
EBA: stripping of (A)
Residue curves Rcm stable separatrix
above (FE/LT)min,A

(c) (d) Chlorform (B)

 
(c) Chlorform (B)  
FE/V = 0.5
    R= ∞
FE/V =0.5
  R= ∞  
 

[UNunstable]
[UNunstable]

XDA

Water(E) Acetone (A)

Acetone (A) Dichloromethane (E)
Extractive profile Extractive profile Extractive separatrix

Figure 7: Thermodynamic feasibility criterion (a, c) for 1.0–1a case a with light
entrainer dichloromethane (b, d) for 3.1–4 case a with water.
Brazilian Journal of Chemical Engineering
 Thermodynamic Topological Analysis of Extractive Distillation of Maximum Boiling Azeotropes 965

other, the relative volatility between A and B is re- high economic and environmental advantage, it is a
versed. Because αA,B=1 crosses the diagram binary limited efficiency solvent for separating the ace-
A-E side, the feasibility criterion is valid for acetone tone/chloroform system. Application of the thermo-
(A) as the lightest component, and there is a residue dynamic criterion hints at product feasibility, in
curve connecting the azeotropic components, ena- which the product goes up or down using a rectifying
bling recovery of acetone as product in a very small or stripping column, operating parameter values and
region AEB. The feasibility criterion also holds for solvent efficiency.
the high-boiling component water (E), enabling its
recovery as product in a stripper. Besides, the separa-
tion using the entrainer water, which is immiscible NOMENCLATURE
with chloroform (see Figure 7(d)), makes the separa-
tion process harder than it should be. Even though A light original component
water as a solvent has a high economic and environ- ABE volatility orders, A is possible distillate
mental advantage, the thermodynamic insight indi- B heavy original component
cates that it is a limited efficiency solvent for sepa- E entrainer
rating the acetone/chloroform system. FE/F entrainer - feed flow rate ratio, continuous
process
FE/V entrainer - feed flow rate ratio, batch
CONCLUSIONS process
(FE/V)min minimum entrainer - feed flow rate ratio,
A topological insight methodology has been pro- batch process
posed to assess the feasibility of the separation of (FE/V)max maximum entrainer - feed flow rate ratio,
aazeotropic mixture by using different kinds of en- batch process
trainer. The possible products in the distillation col- Ki distribution coefficient
umn and the existence of a limiting entrainer/feed RCM Residue Curve Map
flow rate ratio are predicted using general feasibility [SNextr] extractive node feasible range
criteria. The separation of the maximum-boiling aze- TmaxazeAB the maximum azeotrope point of mixture AB
otrope acetone/chloroform is used as a case study, TminazeAB the minimum azeotrope point of mixture AB
using 7 candidate solvents, the ternary system can be maxT the maximum boiling temperature
classified into diagrams class 1.0–1a (using light en- minT the minimum boiling temperature
trainer), class 1.0–2 (using heavy entrainer), and class UN unstable node originating at the entrainer
3.1-4 (using water).The conceptual design of the fea- vertex
sibility study of the three classes is developed by the UN’ unstable node originating outside the
analysis of the residue curve map and product regions. composition simplex
For the class 1.0–2 (separation of acetone/chloro- xp intersection point between univolatility
form by adding heavy entrainer E1-E4), the univola- curve and residue curve passing through
tility line αAB =1 can intersect the B-E edge. The the distillate product
general criterion states that, under infinite reflux
ratio, both A and B can be obtained at the top, de-
pending on the overall feed composition location. REFERENCES
There also exists a maximum value for FE/F when
component A is the distillate. For the class 1.0–1a, Benyahia, K., Benyounes, H. and Shen, W., Energy
corresponding to the separation of a maximum boil- evaluation of ethanol dehydration with glycol
ing azeotropic mixture A-B using a light entrainer mixture as entrainer. Chemical Engineering Tech-
E7, the process is feasible above a minimal limit nology, 37(6), 987-994(2014).
entrainer/feed flow rate ratio value, so that the ex- Benyounes, H., Shen, W. and Gerbaud, V., Entropy
tractive stable node SNextr,A is located in the region flow and energy efficiency analysis of extractive
where it can intersect a stripping profile which can distillation with a heavy entrainer. Industrial En-
reach the expected product A. The possible gineering Chemistry Research, 53(12), 4778-4791
advantage of using a light entrainer is to offer more (2014).
opportunities for separating minT and maxT azeo- Brüggemann, S. and Marquardt, W., Shortcut methods
tropic mixtures when it may not be easy to find a for nonideal multicomponent distillation: 3. Ex-
heavy or intermediate entrainer. The thermodynamic tractive distillation columns. AIChE Journal, 50,
insight indicates that using solvent water corresponds 1129-1149 (2004).
to diagrams class 3.1-4 and, even though it has a Düssel, R. and Stichlmair, J., Separation of azeo-

Brazilian Journal of Chemical Engineering Vol. 32, No. 04, pp. 957 - 966, October - December, 2015
966 W. F. Shen, H. Benyounes and J. Song

tropic mixtures by batch distillation using an 2524-2536 (2002).

entrainer. Computers and Chemical Engineering, Luyben, W. L., Effect of solvent on controllability in
19, 113-118 (1995). extractive distillation. Industrial and Engineering
Gurikov, Y. V., Structure of the vapor–liquid equilib- Chemistry Research, 47, 4425-4439 (2008).
rium diagrams of ternary homogeneous solutions. Luyben, W. L. and Chien, I. L., Design and Control
Russian Journal of Physical Chemistry, 32(9), of Distillation Systems for Separating Azeo-
1980-1996 (1958) (In Russian). tropes. Wiley Online Library (2010).
Hilmen, H. K., Separation of azeotropic mixtures: Matsuyama, H. and Nishimura, H., Topological and
Tools for analysis and studies on batch distillation thermodynamic classification of ternary vapor-
operation. PhD Thesis, Norwegian University of liquid equilibria. Chemical Engineering of Japan,
Science and Technology (2000). 10(3), 181-187 (1977).
Hilmen, E., Kiva, V. and Skogestad, S., Topology of Rodriguez-Donis, I., Gerbaud, V. and Joulia, X.,
ternary VLE diagrams: Elementary cells. AIChE Thermodynamic insights on the feasibility of ho-
Journal, 48,752-759 (2003). mogeneous batch extractive distillation. 1. Azeo-
Jobson, M., Hilderbrandt, D., Glasser, D., Attainable tropic mixtures with a heavy entrainer. Industrial
products for the vapor-liquid separation of ho- Engineering Chemistry Research, 48, 3544-3559
mogenous ternary mixtures. Chemical Engineer- (2009a).
ing Journal, 59, 51-70 (1995). Rodríguez-Donis I., Gerbaud V. and Joulia X., Ther-
Kiva, A. V., Hilmen, E. K. and Skogestad, S., Azeo- modynamic insights on the feasibility of homoge-
tropic phase equilibrium diagrams: A survey. Chem- neous batch extractive distillation. 2. Low-rela-
ical Engineering Science, 58, 1903-1953 (2003). tive-volatility binary mixtures with a heavy en-
Knapp, J. P. and Doherty, M. F., Minimum entrainer trainer. Industrial Engineering Chemistry Re-
- feed flows for extractive distillation: A bifurca- search, 48(7), 3560-3572 (2009b).
tion theoretic approach. AIChE Journal, 40, 243- Rodríguez-Donis, I., Gerbaud, V. and Joulia, X.,
268 (1994). Thermodynamic insights on the feasibility of ho-
Krolikowski, L. J., Distillation regions for non-ideal mogeneous batch extractive distillation. 3. Azeo-
ternary mixtures. International Conference of tropic mixtures with light entrainer. Industrial
Distillation and Absorption (2002). Engineering Chemistry Research, 51, 4643-4660
Lang, P., Yatim, H., Moszkowicz, P. and Otterbein, (2012a).
M., Batch extractive distillation under constant Rodriguez-Donis, I., Gerbaud, V. and Joulia, X.,
reflux ratio. Computer and Chemical Engineer- Thermodynamic insights on the feasibility of ho-
ing, 18, 1057-1069 (1994). mogeneous.batch extractive distillation. 4. Azeo-
Lang, P., Lelkes, Z., Otterbein, M., Benadda, B. and tropic mixtures with intermediate boiling en-
Modla, G., Feasibility studies for batch extractive trainer. Industrial and Engineering Chemistry Re-
distillation with a light entrainer. Computer and search, 51, 6489-6501 (2012b).
Chemical Engineering, 23, 93-96 (1999). Serafimov, L., Thermodynamic and topological anal-
Lang, P., Modla, G., Benadda, B. and Lelkes, Z., ysis of liquid-vapor phase equilibrium diagrams
Homoazeotropic distillation of maximum azeo- and problems of rectification of multicomponent
tropes in a batch rectifier with continuous entrainer mixtures. Mathematical Methods in Contempo-
feeding. I. Feasibility studies. Computers and rary Chemistry (1996).
Chemical Engineering, 24, 1665-1671 (2000a). Shen, W. F., Extension of thermodynamic insights on
Lang, P., Modla, G., Kotai, B., Lelkes, Z. and batch extractive distillation to continuous opera-
Moszkowicz, P., Homoazeotropic distillation of tion. PhD Thesis, Institut National Polytechnique
maximum azeotropes in a batch rectifier with de Toulouse (2012).
continuous entrainer feeding II. Rigorous simula- Shen, W. F., Benyounes, H. and Gerbaud, V., Exten-
tion results. Computers and Chemical Engineer- sion of thermodynamic insights on batch extrac-
ing, 24, 1429-1435 (2000b). tive distillation to continuous operation. 1. Azeo-
Laroche, L., Bekiaris, N., Andersen, H. W. and tropic mixtures with a heavy entrainer. Industrial
Morari, M., The curious behavior of homogene- Engineering Chemistry Research, 52, 4606-4622
ous azeotropic distillation–implications for en- (2013a).
trainer selection. AIChE Journal, 38, 1309-1328 Shen, W. F. and Gerbaud, V., Extension of thermody-
(1992). namic insights on batch extractive distillation to
Lelkes, Z., Rev, E., Steger, C. and Fonyo, Z., Batch continuous operation. 2. Azeotropic mixtures with
extractive distillation of maximal azeotrope with a light entrainer. Industrial Engineering Chemis-
middle boiling entrainer. AIChE Journal, 48, try Research, 52, 4623-4637 (2013b).

Brazilian Journal of Chemical Engineering