Computers and Chemical Engineering 100 (2017) 2737

Contents lists available at ScienceDirect

Computers and Chemical Engineering

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

Comparison of heterogeneous azeotropic distillation and extractive

distillation methods for ternary azeotrope ethanol/toluene/water
separation
Lei Zhao, Xinyu Lyu, Wencheng Wang, Jun Shan, Tao Qiu
Jiangsu Key Laboratory of Advanced Catalytic Materials and Technology, School of Petrochemical Engineering, Changzhou University, No.1, Gehu Road,
Wujin District, Changzhou, Jiangsu 213164, China

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

Article history: Ethanol and toluene are important organic solvents in chemical industry. Ethanol, toluene and water
Received 31 October 2016 present a ternary minimum-boiling azeotrope and three binary azeotropes. In this article, two methods
Received in revised form 1 February 2017 are studied to separate the ternary azeotrope: heterogeneous azeotropic distillation using the toluene,
Accepted 4 February 2017
which is in the system itself, as entrainer and extractive distillation with glycerol as the only one solvent.
Available online 6 February 2017
In the heterogeneous azeotropic distillation, the pressures of the three columns are changed respectively
to achieve heat integration. In the extractive distillation, only one solvent and two columns are used
Keywords:
to separate the ternary mixture. Experiment shows that the partially heat-integrated heterogeneous
Ternary azeotrope
Heterogeneous azeotropic distillation
azeotropic distillation reduces the energy cost and total annual cost by 27.7% and 13.4% respectively
Extractive distillation compared with the conventional heterogeneous azeotropic distillation. Unexpectedly, the extractive dis-
Ethanol/toluene/water tillation can save 18.8% and 39.3% in the energy cost and total annual cost respectively compared with
the partially heat-integrated heterogeneous azeotropic distillation.
2017 Elsevier Ltd. All rights reserved.

1. Introduction rities. This special HAD method is applied to separate azeotrope

Ethanol and toluene are widely used in chemical, pharmaceuti-
cal, dye and other elds as good solvents. Ethanol can deliquesce
easily and absorb moisture from the air quickly because of the pres-
ence of hydrogen bond. Since ethanol, toluene and water present
three binary azeotropes and one ternary azeotrope, ordinary dis-
tillation methods can not separate the mixture effectively. Some
other kinds of distillation methods should be studied to separate
the ethanol/toluene/water mixture.
Many methods can be used to separate azeotropic mixture, for
example, pressure-swing distillation (Yu et al., 2012), azeotropic
distillation (Abu-Eishah and Luyben, 1985), extractive distillation
(Meirelles et al., 1992), adsorption (Garg and Ausikaitis, 1983), per-
vaporation using membrane (Fleming, 1992) and some other new
separation techniques. In this article, heterogeneous azeotropic dis-
tillation (HAD) and extractive distillation (ED) methods are studied
to separate the ethanol/toluene/water mixture.
It is innovative to use toluene, which is in the system itself, as
entrainer for the HAD, which will not bring in any other impu-
Corresponding author.
E-mail address: qiutao@cczu.edu.cn (T. Qiu).
or close-boiling mixtures using the entrainer, which is in the sys-
tem itself, to generate a lower-boiling azeotrope with the one or
two components in the mixture. The minimum-boiling azeotrope
produces two immiscible liquid phases in the decanter. Aque-
ous phase and organic phase both ow back to the column with
optimum reux ratios and the rest are separated and fed into
different columns to obtain pure products. HAD for azeotropic mix-
ture is expected to cross the distillation boundary (Doherty and
Malone, 2001; Widagdo and Seider, 1996). Kiss used HAD method
in bioethanol dehydration with n-pentane as entrainer (Kiss et al.,
2012). Luyben also used benzene as the entrainer in the separation
of ethanol/water azeotrope by HAD (Luyben, 2006a, 2006b).
ED is generally applied to separate binary azeotrope and other
close-boiling mixtures with the relative volatility of light compo-
nents and heavy components below 1.1 (Seader and Henley, 1998;
Errico et al., 2013a, 2013b; Kossack et al., 2008; Lei et al., 2003;
Liang et al., 2014). The solvent added in the mixture improves
the relative volatility of light components and heavy components
because the solvent interacts differently with the light components
and heavy components in the mixture. When the relative volatility
is improved, the light components are acquired in the top of the
extractive column (EC) while the heavy components and solvent
are acquired from the bottom of the EC for subsequent separation.

http://dx.doi.org/10.1016/j.compchemeng.2017.02.007
0098-1354/ 2017 Elsevier Ltd. All rights reserved.
28 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737

Fig. 1. Process ow sheet of HAD.

Table 1
Azeotropic Temperature and Composition of the System at 1 atm.

Temp (K) Type No. Comp. Toluene (mass%) Ethanol (mass%) Water (mass%)

347 Heterogeneous 3 47.49 42.75 9.77

357.68 Heterogeneous 2 80.09 0 19.91
351.3 Homogeneous 2 0 95.62 4.38
350.01 Homogeneous 2 31.92 68.08 0

Solvent and heavy components are separated by a solvent recov-
ery column (SRC) easily. Then, the solvent obtained in the bottom
of SRC is recycled to the EC (Kossack et al., 2008). Luyben used
dimethyl sulfoxide as solvent to separate maximum-boiling ace-
tone/chloroform mixture by ED (Luyben, 2008). Figueroa used ED
method with ionic liquids as solvent to produce anhydrous ethanol
and saved energy a lot (Figueroa et al., 2012). Lei used ED method
with solid inorganic salt and ionic liquid as solvents to separate
ethanol/water mixture (Lei et al., 2014). Long retrotted conven-
tional ED to a thermally coupled distillation scheme and reduced
the energy cost signicantly (Duc Long and Lee, 2013).
Separation of the ethanol/toluene/water mixture has visible
economic benet, social benet and environmental benet. There
are a lot of papers discussing the separation of binary azeotrope
via HAD or ED. However, the articles on the separation of ternary
azeotrope using HAD and ED are quite limited. So, it is neces-
sary to study some workable distillation methods to separate the
ethanol/toluene/water ternary azeotrope, with the emphasis on
economic optimization.
In order to separate the ethanol/toluene/water ternary
azeotrope, two special distillation methods are studied in this arti-
cle: HAD and ED methods. According to the total annual cost (TAC),
optimization programs are used to get optimal congurations of
the two distillation methods.
2. Simulation
2.1. Heterogeneous azeotropic distillation
2.1.1. Process design and ternary diagrams
In the separation of the ternary mixture with
ethanol/toluene/water, the toluene can also serve as an entrainer
and form a low-boiling ternary azeotrope with ethanol and water.
So, it does not need to add any other entrainer, which will not
bring in any other impurities. The ethanol/toluene/water mixture
presents three binary azeotropes and one ternary azeotrope. One
ternary azeotrope (ethanol/toluene/water) is satisfactory because
it is heterogeneous and its azeotropic temperature (347 K) is
the lowest in the system. Another three binary azeotropes are
also formed with azeotropic temperatures of 357.68, 351.3 and
350.01 K, respectively. Table 1 shows the azeotropic temperature
and composition of ethanol/toluene/water mixture at 1 atm. When
the pressure is 2.18 atm or 0.33 atm, the ternary azeotrope is still
heterogeneous and its azeotropic temperature is still the lowest
in the system. The NRTL physical property model is used in the
simulation for the separation of the ternary azeotrope.
 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737
Fig. 2. Residue curve maps for ethanol/toluene/water mixture at (a) 2.18 atm, (b)
0.23 atm, (c) 0.33 atm.
29
Table 2
Goal Functions and Necessary Parameters for Economic Evaluation.
Parameters Data
Condensers Heat transfer
coefcient = 0.852 kW/K m2
Capital cost = 7296 (A, m2 )0.65
Q
Heat transfer area: A = (Kt)
Reboilers Heat transfer
coefcient = 0.568 kW/K m2
Capital cost = 7296 (A, m2 )0.65
Q
Heat transfer area: A = (Kt)
Column vessel Capital
cost = 17640 (D, m)1.066 (L, m)0.802
Length: H = 1.2 0.61 (NT 2)
Energy cost LP stream (433 K): $7.78 per GJ
MP stream (457 K): $8.22 per GJ
HP stream (527 K): $9.88 per GJ
Chilled-water (253.15 K): $7.89 per GJ
totalcapitalcost
TAC paybackperiod
+ annual energy cost
Payback period 3 years
The process ow sheet of the HAD using toluene as entrainer is
shown in Fig. 1 with three columns and one decanter. The pressures
of column 2 (C2 ) and column 3 (C3 ) are both reduced to 0.33 atm to
ensure that the temperature of the condensers is about 323.75 K.
So, cooling water can still be used in the condensers. The pressure
of column 1 (C1 ) is raised to 2.18 atm to ensure that the tempera-
ture in the top stage of C1 is about 20 K more than the temperature
in the bottom of C2 and C3 . So, heat integration can be achieved in
the HAD in the next section. The C1 has a decanter and the feed-
stock is fed into this column (Stage 1 is the top tray). High-purity
ethanol product is obtained from the bottom of C1 while the min-
imum boiling ethanol/toluene/water heterogeneous azeotrope is
obtained from the top of C1 . Then, the overhead vapor stream of
C1 is cooled, and two liquid phases are separated in the decanter.
Some of both the aqueous and organic phases are reuxed to the
top tray. The rest of the aqueous phase is fed into C2 and the rest
of the organic phase is fed into C3 . High-purity water and toluene
products are obtained in the bottom of C2 and C3 respectively while
the top products of these two columns are combined and back to
C1 .
The residue curve map (RCM) of ethanol/toluene/water mixture
is shown in Fig. 2, which is divided into three distillation areas by
the distillation boundaries. The critical regions are enlarged in the
ternary diagrams. Fig. 2a is with a pressure of 2.18 atm, the oper-
ating pressure in C1 . Because the temperature in the decanter is
313.15 K, Fig. 2b is with a pressure of 0.23 atm, where the temper-
ature of the ternary azeotrope is 313.59 K (Luyben, 2006a, 2006b).
Fig. 2c is with a pressure of 0.33 atm, the operating pressure in C2
and C3 . Fig. 2 also shows the azeotropic temperature and compo-
sition of ethanol/toluene/water mixture at 2.18 atm and 0.33 atm.
Distillation boundary represents the limit of the separation in a
column. The composition points of overhead product and bot-
tom product in one column must lie in the same distillation area
(Luyben, 2006a, 2006b). The ethanol product is near the top corner
in area 1. The water product is near the lower left corner in area 2.
The toluene product is near the lower right corner in area 3. This
means that three columns at least must be used for the desired
separation.
In Fig. 2a, the composition points in C1 are analyzed without
the decanter. The system has four inputs: F1 , FR, aqueous reux
(AR) and organic reux (OR), and it has two outputs: B1 and V. M
(F1 + FR + AR + OR), B1 and V composition points lie on the straight
line and in the same area of the RCM, so M can be separated in one
column and the ethanol product can be obtained in the bottom of
C1 . Because Fig. 2a is with a pressure of 2.18 atm, where the temper-
ature of the ternary azeotrope is 368.41 K, not the 313.15 K in the
30 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737

Fix P1=2.18 atm P2=0.33 atm

P3=0.33 atm

Give NT3

Give NF3 No
No

Give NT2 No

Give NF2

Is TAC minimal with

Give NT1
NT3, NF3?

Give NFR, NF1

Yes

Vary RRa, RRo, RR2, RR3 to meet

design specifications of column C1, C2 Is TAC minimal with
No NT3?
and C3
No
Yes
No Is TAC minimal with NT1,
NT2, NF2, NT3, NF3?

Is TAC minimal?
Yes

Is TAC minimal with NT2, YES Yes

NF2, NT3, NF3?
Get the optimal NT1, NT2, NT3, NF1,
NFR, NF2, NF3 and TAC
Yes

Is TAC minimal with NT2,

Over
NT3, NF3?

Fig. 3. Sequential iterative optimization procedure for HAD.

decanter, AR and OR composition points are not on the liquidliquid
equilibrium envelope.
In Fig. 2b, AR + F2 and OR + F3 composition points will lie on a
straight line and pass through V composition point. However, the
three composition points are in different areas of the RCM. The
decanter permits AR + F2 , OR + F3 and V composition points to cross
the distillation boundary. In addition, AR + F2 and OR + F3 composi-
tion points are on the liquidliquid equilibrium envelope. Because
Fig. 2b is with a pressure of 0.23 atm, V composition point is not in
area 1.
Fig. 2c shows the composition points in C2 and C3 . F2 is separated
into B2 and D2 . F3 is separated into B3 and D3 . F2 , B2 and D2 com-
position points lie on the straight line and in the same area of the
RCM, so F2 can be separated in one column and water product can
be obtained in the bottom of C2 . F3 , B3 and D3 composition points
lie on a straight line and in the same area of the RCM, so F3 can be
separated in one column and toluene product can be obtained in
C3 . This means that the three products can be obtained using three
columns and one decanter.
2.1.2. Process study and economic analysis
The following data is used for the distillation process
simulation: the feedstock consists of 61.52/32.97/5.51 mass%
ethanol/toluene/water, with the ow rate 1000 kg/h. The three
product specications are set as follows: the ethanol product has a
purity of 99.9 mass%, the toluene product has a purity of 99.9 mass%
and the water product has a purity of 99.5 mass% by varying reux
ratio of aqueous phase (RRa) and reux ratio of organic phase (RRo)
in C1 , and reux ratios in C2 and C3 (RR2 , RR3 ) at the same time. The
assumed tray pressure drop of each column is 0.0068 atm. The dis-
tillation process is simulated by Aspen Plus. Luybens book explains
how to use the software expertly (Luyben, 2013).
In this article, the reux drum is regarded as the rst stage
and the reboiler is regarded as the last stage (Luyben, 2009; Yang
et al., 2013). TAC is minimized in design of the distillation to get
a better process. TAC is minimized by changing the number of
stages and the feed stage, etc. to get the optimum design. Con-
ventionally, TAC consists of process variable costs and xed capital
investment. Because the price of the decanter, reux drum, pumps,
valves and pipes is usually far less than the price of the towers
and heat exchangers, these equipments are usually not considered.
The most important equipments include column vessels, reboilers
and condensers. Table 2 shows the goal functions for the economic
evaluation (Luyben, 2006a, 2006b).
In C1 , the variables contain the number of stages (NT1 ), fresh feed
stage (NF1 ) and recycled stream feed stage (NFR ). In C2 , the variables
 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737 31

(a) 3.770 (b) 3.81

3.765 3.80

3.79
3.760

3.78
TAC(10 $/y)

TAC(10 $/y)
3.755
5

3.77

5
3.750
3.76

3.745
3.75

3.740 3.74

3.735 3.73
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 10 11 12 13 14 15 16 17 18 19 20 21 22
NT1(the column C1) NT2(the column C2)

(c) 3.755

3.750

3.745
TAC(10 $/y)

3.740
5

3.735

3.730

3.725

3.720
2 3 4 5 6 7 8 9 10 11
NT3(the column C3)

Fig. 4. TAC with stages of (a) C1 , (b) C2 and (c) C3 .

contain the number of stages (NT2 ) and feed stage (NF2 ). In C3 , the and C3 with energy. Two auxiliary condensers are added in C1 to
variables contain the number of stages (NT3 ), and feed stage (NF3 ). replenish necessary condensation.
These seven variables are optimized in order to minimize TAC of The optimal congurations for the partially heat-integrated
the HAD. An economic calculation sequence is set up to optimize HAD are found in the conventional HAD. The energy cost of the
the economics when the system has so many variables. Fig. 3 shows partially heat-integrated HAD is $1.54 105 /y, and the TAC is
the sequential iterative optimization procedure expressly. $3.23 105 /y. Optimal ow sheet for partially heat-integrated HAD
Fig. 4a shows that the optimal NT1 is 75 with optimum NFR with equipment data, steam data, reux ratio, heat duties, internal
and NF1 . Fig. 4b shows that the optimal NT2 is 15 with optimum diameter (ID) and operating pressures (P), etc. is shown in Fig. 6.
NF2 . Fig. 4c shows that the optimal NT3 is 5 with optimum NF3 . Comparing the two optimal HAD methods, the research indi-
The optimal economic parameters are obtained: the optimal design cates that partially heat-integrated HAD is more economical than
variables are NT1 = 75, NT2 = 15 and NT3 = 5. The energy cost of the conventional HAD. Experiment shows that the partially heat-
HAD is $2.13 105 /y, and the TAC is $3.73 105 /y. Optimal ow integrated HAD offers 27.7% and 13.4% reduction in the energy cost
sheet for HAD with equipment data, steam data, reux ratio, heat and TAC respectively.
duties, internal diameter (ID) and operating pressures (P), etc. is
shown in Fig. 5. 2.2. Extractive distillation

2.1.3. Partially heat-integrated heterogeneous azeotropic 2.2.1. Solvent selection

distillation More time should be spent selecting the potential solvents
Many scientists have made great efforts to increase the ef- because the solvent plays a vital role in ED.
ciency of energy utilization for the distillation methods in the past The solvent should be less volatile than components in the
decades. It is generally known that the columns of the pressure- mixture. The solvent should increase the relative volatility of the
swing distillation are set at different pressures. So, the energy components in the mixture and not generate an azeotrope with
utilization efciency of the pressure-swing distillation can be any component in the mixture. Also, the solvent should be readily
increased by heat integration. Different operating pressures can available and cheap.
also be applied to HAD for heat integration. Originally, two solvents should be used to increase the relative
The ow sheet for partially heat-integrated HAD is similar to volatility of ethanol/toluene and ethanol/water respectively. How-
the conventional HAD mentioned above. In order to achieve heat ever, experiment shows that only one solvent is needed to increase
integration, the overhead vapor from C1 provides reboilers of C2 the relative volatility of ethanol/toluene and ethanol/water respec-
32 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737

Fig. 5. Optimal ow sheet for HAD.

Fig. 6. Optimal ow sheet for partially heat-integrated HAD.

 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737 33

(a) 1.0 (b) 1.0

0.9 0.9

0.8 0.8

0.7 0.7

0.6 0.6

y(Ethanol)
y(Ethanol)

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2
ethanol-toluene-glycerol ethanol-water-glycerol
0.1 ethanol-toluene 0.1 ethanol-water
0.0 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
x(Ethanol) x(Ethanol)

Fig. 7. xy diagram for (a) ethanol + toluene + glycerol, (b) ethanol + water + glycerol system.

Fig. 8. Process ow sheet of ED.

Fig. 9. Effect of S and RR1 in the EC (NT1 = 45, NF1 = 26, NFE = 5) on (a) ethanol purity in D1 and (b) impurity of ethanol in B1 .
34 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737
Fig. 10. Sequential iterative optimization procedure for ED.
Table 3
Results of the Solvent Selection for ED by Aspen Plus.
solvent Tb /K
glycerol 564.05
ethylene glycol 470.45
n-propylbenzene 432.35
n-butylbenzene 455.25
tively. Four solvents are discussed in this article for the separation
of ethanol/toluene/water mixture: glycerol (Gil et al., 2012), ethy-
lene glycol, n-propylbenzene and n-butylbenzene. Table 3 shows
their boiling point (Tb) and the relative volatility (1,2 , 1,3 the
relative volatility of ethanol/toluene and ethanol/water with sol-
vent S). Glycerol is chosen as the solvent in the simulation because
glycerol does not bring in any other azeotropes in the mixture
and can increase the relative volatility of ethanol/toluene (5.19)
and ethanol/water (5.82) simultaneously. Fig. 7 shows that glycerol
modies the vapor-liquid equilibrium curve obviously and elimi-
nates ethanol/water and ethanol/toluene azeotrope. The effect of
glycerol allows obtaining high purity ethanol. So, only one sol-
vent will be used in the ED. This characteristic will bring in fewer
impurities and greatly reduce the energy cost.
1,2 (ethanol/toluene) 1,3 (ethanol/water)
5.19 5.82
0.296 4.9
2.166 0.196
2.128 0.321
2.2.2. Process design and economic analysis
More attention should be paid to the process design and eco-
nomic analysis after glycerol is selected as the solvent. Only two
columns will be used in the ED thanks to the immiscibility of
toluene with water. The energy cost and xed capital investment
will be reduced signicantly because one column is cut down. This
is infrequent in the separation of ternary azeotrope. The ED process
with two columns and one decanter is shown in Fig. 8 clearly. The
fresh stream is fed at the middle stage of the EC while the solvent is
fed at the top stage of the EC. In the EC, glycerol increases the volatil-
ity of ethanol/water and ethanol/toluene respectively and makes
the separation easier. Glycerol, toluene and water ow down and
outow in the bottom of the EC while high purity ethanol prod-
uct is obtained from the top of the EC. The SRC has a decanter and
removes water and toluene from glycerol. This separation is very
 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737 35

(a) 2.05 (b) 2.000

2.04 1.995

2.03 1.990

2.02 1.985
TAC(10 $/y)

TAC(10 $/y)
2.01 1.980

5
5

2.00 1.975

1.99 1.970

1.98 1.965

1.97 1.960

1.96 1.955
1.2 1.3 1.4 1.5 1.6 1.7 1.8 37 38 39 40 41 42 43 44 45 46 47

solvent ratio(S/F) NT1(the extractive column)

(c) 2.05

2.04

2.03

2.02

2.01
TAC(10 $/y)
5

2.00

1.99

1.98

1.97

1.96

1.95
2 3 4 5 6 7 8
NT2(the solvent recovery column)

Fig. 11. TAC with (a) S/F ratio (mass), stages of (b) the EC (c) SRC.

easy because water and toluene is much more volatile than glyc-
erol. The overhead vapor stream of toluene/water is cooled, and two
liquid phases are separated in the decanter because of the immisci-
bility of toluene with water. Some of both the aqueous and organic
phases are reuxed to the top tray. The rest of the aqueous phase
and organic phase are water and toluene products. Then, the sol-
vent is cooled and recycled to the EC. Finally, little pure solvent
make-up is added to maintain the solvent ow rate/fresh feed ow
rate (S/F) ratio constant.
Three product specications are set as follows: the ethanol prod-
uct has a purity of 99.9 mass%, the toluene product has a purity of
99.9 mass% and the water product has a purity of 99.5 mass%. The
pressure of the EC is operated at 1 atm. The pressure of SRC is oper-
ated at 0.05 atm to avoid the thermal degradation of glycerol due to
the high temperature (Gil et al., 2012). The assumed tray pressure
drop of each column is 0.0068 atm.
There are two design degrees for the EC when the number of
stages (NT1 ), fresh feed stage (NF1 ), solvent feed stage (NFE ) and
operating pressure (P1 ) are given: solvent ow rate (S) and reux
ratio (RR1 ). Much time is spent exploring the effect of S and RR1
on compositions of overhead stream and bottom stream before
the strict simulation. The effect of S and RR1 on the compositions
of overhead stream and bottom stream in the EC with NT1 = 45,
NF1 = 26, NFE = 5 is shown in Fig. 9. The effect of S and RR1 on the
impurity of ethanol in B1 is shown in Fig. 9b. There will be less
ethanol in B1 if S could be higher. In order to acquire the 99.5 mass%
water product in SRC, the mass ow rate of ethanol in B1 must be
less than 0.316 kg/h. So, the S must exceed 1280 kg/h.
Fig. 9a shows that an optimal RR1 makes the maximum ethanol
purity with a given S. The higher the S is, the higher the ethanol
purity could be obtained. Fig. 9a shows that the purity of ethanol
could be more than 99.9% with an optimum RR1 when S is
1280 kg/h.
In the EC, the bottom stream is held at 0.316 kg/h ethanol and
the distillate stream is held at 99.9 mass% ethanol by varying reux
ratio (RR1 ) and distillate ow rate (D1 ). In the SRC, the bottoms
stream is held at 99.9 mass% glycerol by varying reux ratio of
aqueous phase (RRa) and reux ratio of organic phase (RRO ). The
distillate stream is held at 0.01 mass% glycerol by vary distillate
ow rate (D2 ).
A cooler is required to cool the recycled solvent. Knight and
Doherty suggested that the temperature of solvent stream should
be 515 K less than the temperature in the top of the EC (Knight
and Doherty, 1989). So, the temperature of the solvent stream is
set at 336.15 K.
In the EC, the variables contain the number of stages (NT1 ), fresh
feed stage (NF1 ), and solvent feed stage (NFE ). In the SRC, the vari-
ables contain the number of stages (NT2 ) and feed stage (NF2 ). These
ve variables are optimized to make TAC minimal for the ED. An
economic calculation sequence is set up to optimize the economics
36 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737

Fig. 12. Optimal ow sheet for ED.

Table 4
Optimal Results for Partially Heat-Integrated HAD and ED.

Parameter Partially Heat-Integrated HAD ED

C1 C2 C3 EC SRC

NT 75 15 5 41 3
NF 5 9 2 25 2
NFR /NFE 6 N/A N/A 3 N/A
RR 1.72 (RRa)1.72 (RRo) 0.37 1.00 0.27 0.053 (RRa)0.053 (RRo)
Reboiler heat (kW) 735.78 245.78 36.05 346.85 108.31
Total capital cost (105 $) 5.07 2.14
Total energy cost (105 $/y) 1.54 1.25
TAC (105 $/y) 3.23 1.96

when the system has so many variables. Fig. 10 shows the sequen-
tial iterative optimization procedure expressly. QR1 is the reboiler
heat for the EC. QR2 is the reboiler heat for the SRC.
Fig. 11a shows that the optimal S/F ratio is 1.5. Fig. 11b shows
that the optimal NT1 is 41 with optimum NFE and NF1 . Fig. 11c
shows that the optimal NT2 is 3 with optimum NF2 . The optimal
economic parameters are obtained: the optimal design variables
are NT1 = 41 and NT2 = 3. The energy cost of the ED is $1.25 105 /y,
and the TAC is $1.96 105 /y. Optimal ow sheet for ED with equip-
ment data, steam data, reux ratio, heat duties, internal diameter
(ID) and operating pressures (P), etc. is shown in Fig. 12. It is impor-
tant to observe that the temperature of the condenser in the SRC is
281.41 K. More expensive chilled-water should be used to condense
the overhead vapor stream.
3. Process comparison
In this chapter, we will compare partially heat-integrated HAD
with ED according to the economics. The optimal processes of the
two kinds of special distillation are acquired by the same standard
and evaluation procedures. Table 4 shows the optimal result of par-
tially heat-integrated HAD and ED. It is worth mentioning that the
ED has reduced energy cost and TAC by 18.8% and 39.3% respectively
compared with the partially heat-integrated HAD. This is mainly
because only two columns and one solvent are used to separate
the ternary mixture in the ED. Also, evaporation capacity in ED is
less than evaporation capacity in HAD signicantly. It means that
the xed capital investment and energy cost will be reduced a lot
respectively.
4. Conclusion
Two methods for the separation of ethanol/toluene/water
azeotrope are investigated: partially heat-integrated heteroge-
neous azeotropic distillation and extractive distillation. The two
distillation methods are optimized by a proposed optimization
method with TAC as the objective function.
In the partially heat-integrated heterogeneous azeotropic dis-
tillation, the optimal number of stages is 75 for C1 , 15 for C2 and
5 for C3 . The pressures of C1 , C2 and C3 are 2.18 atm, 0.33 atm and
0.33 atm respectively.
Glycerol is chosen as the suitable solvent for the extractive dis-
tillation. The optimal number of stages is 41 for the EC and 3 for the
SRC. The pressures of EC and SRC are 1 atm and 0.05 atm respec-
tively.
 L. Zhao et al. / Computers and Chemical Engineering 100 (2017) 2737
The partially heat-integrated heterogeneous azeotropic dis-
tillation has reduced the energy cost and TAC by 27.7% and
13.4% respectively compared with the conventional heterogeneous
azeotropic distillation. The extractive distillation has reduced
the energy cost and TAC by 18.8% and 39.3% respectively com-
pared with the partially heat-integrated heterogeneous azeotropic
distillation. It is concluded that the extractive distillation is
more attractive than the partially heat-integrated heterogeneous
azeotropic distillation in terms of economics.
Acknowledgments
We are thankful for assistance from the staff at Jiangsu Key Labo-
ratory of Advanced Catalytic Materials and Technology (Changzhou
University). This research did not receive any specic grant from
funding agencies in the public, commercial, or not-for-prot sec-
tors.
References
Abu-Eishah, S.I., Luyben, W.L., 1985. Design and control of a two-column
azeotropic distillation system. Ind. Eng. Chem. Process Des. Dev. 24, 132140.
Doherty, M.F., Malone, M.F., 2001. Conceptual Design of Distillation Systems.
McGraw-Hill Science/Engineering/Math.
Duc Long, N., Lee, M., 2013. Optimal retrot design of extractive distillation to
energy efcient thermally coupled distillation scheme. AIChE J. 59, 11751182.
Errico, M., Rong, B.-G., Tola, G., Spano, M., 2013a. Optimal synthesis of distillation
systems for bioethanol separation. Part 1: extractive distillation with simple
columns. Ind. Eng. Chem. Res. 52, 16121619.
Errico, M., Rong, B.-G., Tola, G., Spano, M., 2013b. Optimal synthesis of distillation
systems for bioethanol separation. Part 2. Extractive distillation with complex
columns. Ind. Eng. Chem. Res. 52, 16201626.
Figueroa, J., Lunelli, B.H., Maciel Filho, R., Maciel, M.W., 2012. Improvements on
anhydrous ethanol production by extractive distillation using ionic liquid as
solvent. Procedia Eng. 42, 10161026.
Fleming, H.L., 1992. Consider membrane pervaporation. Chem. Eng. Prog. 88,
4652.
Garg, D., Ausikaitis, J., 1983. Molecular-sieve dehydration cycle for high
water-content streams. Chem. Eng. Prog. 79, 6065.
37
Gil, I.D., Gmez, J.M., Rodrguez, G., 2012. Control of an extractive distillation
process to dehydrate ethanol using glycerol as entrainer. Comput. Chem. Eng.
39, 129142.
Kiss, A.A., David, J., Suszwalak, P., 2012. Enhanced bioethanol dehydration by
extractive and azeotropic distillation in dividing-wall columns. Sep. Purif.
Technol. 86, 7078.
Knight, J.R., Doherty, M.F., 1989. Optimal design and synthesis of homogeneous
azeotropic distillation sequences. Ind. Eng. Chem. Res. 28, 564572.
Kossack, S., Kraemer, K., Gani, R., Marquardt, W., 2008. A systematic synthesis
framework for extractive distillation processes. Chem. Eng. Res. Des. 86,
781792.
Lei, Z., Li, C., Chen, B., 2003. Extractive distillation: a review. Sep. Purif. Rev. 32,
121213.
Lei, Z., Xi, X., Dai, C., Zhu, J., Chen, B., 2014. Extractive distillation with the mixture
of ionic liquid and solid inorganic salt as entrainers. AIChE J. 60, 29943004.
Liang, K., Li, W., Luo, H., Xia, M., Xu, C., 2014. Energy-efcient extractive distillation
process by combining preconcentration column and entrainer recovery
column. Ind. Eng. Chem. Res. 53, 71217131.
Luyben, W.L., 2006a. Distillation Economic Optimization. John Wiley & Sons, Inc,
Chapter 4.
Luyben, W.L., 2006b. Control of a multiunit heterogeneous azeotropic distillation
process. AIChE J. 52, 623637.
Luyben, W.L., 2008. Control of the maximum-boiling acetone/chloroform
azeotropic distillation system. Ind. Eng. Chem. Res. 47, 61406149.
Luyben, W.L., 2009. Design and control of the monoisopropylamine process. Ind.
Eng. Chem. Res. 48, 1055110563.
Luyben, W.L., 2013. Distillation Design and Control Using Aspen Simulation. John
Wiley & Sons.
Meirelles, A., Weiss, S., Herfurth, H., 1992. Ethanol dehydration by extractive
distillation. J. Chem. Technol. Biotechnol. 53, 181188.
Seader, J.D., Henley, E.J., 1998. Separation Process Principles. John Wiley & Sons,
New York, pp. 604.
Widagdo, S., Seider, W.D., 1996. Journal review. azeotropic distillation. AIChE J. 42,
96130.
Yang, S., Wang, Y., Bai, G., Zhu, Y., 2013. Design and control of an extractive
distillation system for benzene/acetonitrile separation using dimethyl
sulfoxide as an entrainer. Ind. Eng. Chem. Res. 52, 1310213112.
Yu, B., Wang, Q., Xu, C., 2012. Design and control of distillation system for
methylal/methanol separation. Part 2: pressure swing distillation with full
heat integration. Ind. Eng. Chem. Res. 51, 12931310.