Chemical Engineering Science 135 (2015) 166–178

Contents lists available at ScienceDirect

Chemical Engineering Science

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

Design/optimization of energy-saving extractive distillation process

by combining preconcentration column and extractive
distillation column
Yi An, Weisong Li, Ye Li, Shanyuan Huang, Jian Ma, Changlin Shen, Chunjian Xu n
State Key Laboratory of Chemical Engineering, Co-innovation Center of Chemistry and Chemical Engineering, Tianjin University, Tianjin 300072, China

H I G H L I G H T S G R A P H I C A L A B S T R A C T

An innovative two-column extrac-

tive distillation process was devel-
oped.

The new process is energy efficient
with lower capital investment.

Two systems were investigated to
verify the energy and economic
advantages.
Path from conventional setup to energy-saving new process.

art ic l e i nf o a b s t r a c t

Article history: In extractive distillation, the addition of entrainer may reverse the relative volatility of the feed
Received 29 January 2015 components causing the component with higher-boiling point to be removed as the distillate from the
Received in revised form extractive column. In this study, the energy saving possibility of a three-column extractive distillation
25 April 2015
system was studied when this phenomenon occurred. Based on a three-column conventional extractive
Accepted 2 May 2015
Available online 9 May 2015
distillation system, an innovative energy-saving extractive distillation process with lower capital
investment was developed by combining preconcentration column and extractive distillation column.
Keywords: Two important case studies were investigated to verify the above-mentioned energy and economic
Extractive distillation advantages: the separation of n-propanol-water using N-methyl-2-pyrrolidone (NMP) as entrainer and
Energy-saving
the separation of ethyl acetate–ethanol using furfural as entrainer. In order to separate these two
Process design
mixtures, first a three-column extractive distillation sequence including a preconcentration column was
applied to diluted fresh feedstock. Then, based on this three-column conventional extractive distillation
system, an innovative energy-saving distillation process was developed. For the two separated mixtures,
based on the method of global economic optimization, a scheme with optimum design variables was
developed for both of the conventional distillation and the new proposed distillation processes. Results
revealed that the new process offered 22.76% and 17.25% energy-savings respectively. Similar percentage
of reduction in total annual costs (TAC) can also be obtained when compared to the conventional
distillation process.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction difference of the components to achieve desired separation.

Distillation is by far the most important separation process in
chemical engineering, and it exploits the relative volatility
n
Corresponding author. Tel./fax: þ 86 22 27404440.
E-mail address: cjxu@tju.edu.cn (C. Xu).
However, for systems with close boiling point or azeotropic
systems, a separation by conventional distillation process becomes
difficult or even impossible. For these systems, several nonconven-
tional techniques such as heterogeneous azeotropic distillation,
pressure-swing distillation, and extractive distillation have been
proposed (Doherty and Malone, 2001; Seader and Henley, 1998).
However, the pressure-swing option is not economically feasible if

http://dx.doi.org/10.1016/j.ces.2015.05.003
0009-2509/& 2015 Elsevier Ltd. All rights reserved.
 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178
the azeotropic composition is not sufficiently sensitive to press-
ure, and some complexities of dealing heterogeneous azeotropic
distillation, such as multiple steady states, parametric sensitivity
that were clearly illustrated by many authors which consequently
make the design and control challenging (Doherty and Malone,
2001; Seader and Henley, 1998; Widagdo and Seider, 1996).
Extractive distillation is the most widely used form of homo-
geneous azeotropic distillation due to the low energy consumption
and flexible selection of the possible entrainers (Lei et al., 2003;
Sucksmith, 1982). In extractive distillation, a heavier entrainer that
has different interaction with the azeotropic components is added
to alter the relative volatility of the components, thereby effec-
tively “breaking” the azeotrope. Nowadays, extractive distillation
with ionic liquids or the mixture of ionic liquids and solid
inorganic salt as entrainers attracted more and more attention
(Lei et al., 2014a, 2014b).
An important and special aspect of extractive distillation is that
we are not free to pick which of the components in the feed will be
the distillate from the extractive column. For a given entrainer, one
and only one of the feed components can be recovered in the
distillate from the extractive column, and it is not always the
component with the lowest boiling point. For example, the separa-
tion of methanol and acetone using chlorobenzene causes higher-
boiling methanol to be recovered as the distillate from the extractive
column, and the lower-boiling acetone to leave in the bottom stream
with chlorobenzene. Other entrainers, such as water, dimethyl
sulfoxide (DMSO) causes acetone to be recovered as the distillate
from the extractive column (Kossack et al., 2008; Luyben, 2008a,
2008b; Kotai et al., 2007; Modla and Lang, 2012, 2011; Langston et
al., 2005; Gil et al., 2009). It is important to know which of the feed
components will appear in the distillate stream from the extractive
column in order to design the equipment. Isovolatility curve and
pseudo-binary (entrainer-free) equilibrium diagram are reliable
ways. (Doherty and Malone, 2001; Hsu et al., 2010).
Extractive distillation is widely used in a number of processes
(Lladosa et al., 2011; Pacheco-Basulto et al., 2012; Arifin and Chien,
2008; Abushwireb et al., 2007; Wang et al., 2012; Gil et al., 2012; Xu
and Wang, 2006; Hsu et al., 2010), but it is still energy intensive,
different methods to overcome the drawback of energy intensive
have been reported. Process integration is a useful option and it has
been proven that the process integration leads to a significant
reduction in energy consumption compared to conventional process
with no integration. Doherty et al. (Knapp and Doherty, 1990)
developed thermally-integrated extractive distillation sequences for
the separation of ethanol from water using ethylene glycol as
MAKEUP
D1
F
A+B
B1
Preconcentration Extractive distillation
column column
Fig. 1. Sketch of conventional extractive distillation process.
167
entrainer and methanol from acetone using water as entrainer,
results showed that thermally-integrated extractive distillation sys-
tem greatly reduced the energy consumption. Abushwireb et al.
(2007) investigated the recovery of aromatics from pyrolysis gaso-
line using N-methylpyrolidone as entrainer, the results presented in
their report proved that the heat-integrated extractive distillation
system was the best candidate. Luyben (2008b) also studied the
heat-integrated extractive distillation system for the separation of
acetone–methanol using water as entrainer, 27% reboiler duty was
saved as compared to the two-column sequence without thermal
integration. In recent years, the use of dividing wall columns has
drawn considerable attention. Kiss and Suszwalak (2012) studied
bioethanol dehydration by extractive dividing-wall column (E-DWC)
system, which led to 10% energy savings. In our previous study, the
E-DWC system was used to separate methylal–methanol mixture,
results showed that 8% reboiler duty could be saved (Xia et al., 2012).
Wu et al. (2013) made a critical assessment of the energy-saving
potential of E-DWC, they found that the energy-saving potential of
E-DWC was limited.
Although the methods mentioned above are energy efficient,
they have some defects along the way. For example, heat-integrated
extractive distillation systems lead to high investment cost (Knapp
and Doherty, 1990). E-DWC has not been widely used in industry yet
due to the complexities of design and control. As stated by Errico
et al. (2013a, 2013b, 2013c), many studies just focus on extractive
distillation column and entrainer recovery column without consid-
ering the preconcentration column. The preconcentration column of
the process, which in fact is commonly used in extractive distillation
system especially for diluted fresh feed should also be considered
since they are also energy intensive.
This paper intends to investigate the energy-saving possibility of
extractive distillation system when the adding of the high-boiling
solvent reverse relative volatility of feed components. Based on a
three-column conventional extractive distillation system, an inno-
vative extractive distillation process will be developed. Two case
studies will be used as demonstrating examples to verify the energy
and economic advantages of the new process. In the studies, based
on global economic optimization procedure, an optimum design for
both conventional and innovative process will be investigated.
2. Problem statement and process synthesis procedure
In general, for diluted two-component mixtures, the extractive
distillation process is made up of three columns: preconcentration
D3
D2
B A
B2 B3
Entrainer recovery
column
168 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178
distillation column (PDC), extractive distillation column (EDC) and
entrainer recovery column (ERC). The process flow diagram of the
extractive distillation process is presented in Fig. 1. The diluted
mixture A–B is fed into the PDC to separate a pure B (higher-
boiling point) as bottom product and a near-azeotropic composi-
tion mixture of A–B as distillate product. Then, the distillate
product taken from the PDC is sent afterward to the second
column (EDC). In the EDC unit, the extractive entrainer S is added
near the top of the EDC. Due to the addition of entrainer the
relative volatility of A and B is changed, thereby causing the
component having the greater volatility, not necessarily the
component with the lowest boiling point, is taken overhead as a
pure distillate. The other feed component and S are recovered in
the bottom of the extractive distillation column and then fed to
the ERC. The heavy entrainer S is separated from the remaining
component in ERC and then recycled to the EDC. A small makeup
stream of entrainer is added to balance tiny entrainer losses in
both distillate streams.
As shown in Fig. 1, it is worth noting that two pure B streams
are obtained from the bottom of the PDC and the top of the EDC
respectively, which is an opportunity to improve the energy
efficiency. Instead of cooling the overhead vapor of the EDC to
serve as the liquid reflux flow, the vapor flow can be directly fed to
the bottom of the PDC, which provides part of the vapor required
for the PDC, thereby causing the vapor flow generated by the
reboiler of the PDC to be reduced. Due to the reduction of the
vapor flow provided by the reboiler of the PDC, the reboiler heat
duty and heat transfer area can then be significantly decreased.
Part of liquid B stream can be directly fed on the top of the EDC,
thus serving as the liquid-phase flow.
The process flow diagram of the new sequence is presented in
Fig. 2 including two distillation units: combined preconcentration/
extractive distillation column, entrainer recovery column. The first
column in the sequence can be divided into two sections: the
preconcentration section and extractive distillation section. Its
function corresponds to that of the preconcentration column and
A+B Preconcentration
section
B A
MAKEUP
S Extractive
distillation
section
B1
A+S
Combined
preconcentration/extractive
column
Fig. 2. Sketch of combined preconcentration/extractive distillation process.
the extractive distillation column in the conventional extractive
distillation system. The preconcentration section and extractive
distillation section is in fact acting as the PDC and EDC unit of the
conventional sequence respectively.
The diluted mixture A–B is fed into the preconcentration
section, and a near-azeotropic composition mixture of A–B is
removed as distillate. Then, the distillate product taken from the
preconcentration section is sent afterward to the extractive
distillation section. The feed component A and S are recovered in
the bottom of the first column and then fed to the ERC. The heavy
entrainer S is separated from component A in ERC and then
recycled to the first column. As shown in Fig. 2, there is only one
product B stream rather than two products B stream in Fig. 1, and
it is removed as liquid side stream of the first column. It is worth
noting that the intermediate heat exchange is an option. If the
vapor flow provided by the extractive distillation section is not
sufficient, a sidestream reboiler has to be installed to return the
required amount of vapor flow to the preconcentration section.
Otherwise, the sidestream reboiler is not required.
Hence, a two-column extractive distillation system is generated
by combining preconcentration/extractive distillation columns
which is energy efficient as well as has lower capital investment.
In this paper, two case studies are studied to testify the above-
mentioned energy and economic advantages: the separation of n-
propanol/water using N-methyl-2-pyrrolidone (NMP) as entrainer,
and the separation of ethyl acetate/ethanol using furfural as
entrainer.
3. Process design and economical optimization
The new process is more complicated when compared with the
conventional extractive distillation process. Also from the analysis
above, the conventional extractive distillation process can be
directly retrofitted to the new process. Therefore, the steady state
design begins from the conventional extractive distillation system.
D1
D2
B2
Entrainer
recovery
column
 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178
For the new process, the first column is a special column with
complex configuration. From Figs. 1 and 2 the path from conven-
tional process to the new process, it is clear to know that when the
combined preconcentration/extractive distillation column is
decomposed, it should be exactly the PDC and EDC in the
conventional extractive distillation system. Thus, after we get the
optimal design variables of the conventional extractive distillation
system, the optimum design variables of the conventional extrac-
tive distillation system can be directly applied to the new process.
Therefore, the original optimal configurations are all kept the same
in the new process (Errico et al., 2014a, 2014b).
3.1. Process design
In this work, the commercial software Aspen Plus V7.1 was
used in the steady state simulations. The desired product quality
requirements were achieved by the “Design Spec/Vary” feature in
Aspen Plus. In all cases, the reflux ratio and distillate rate were
used as the operation variables of each column.
The PDC and the ERC are all simple columns with one feed
stream and two product streams. It's easy to design these two
columns. The EDC with two feed streams and two product streams,
it is more complicated. There are some very simple and reliable
tools can be used. Residue curve maps that allow us to determine
whether the entrainer is feasible and they can be used as simple
method to analysis and design feasible extractive distillation
sequence (Doherty and Malone, 2001; Doherty and Caldarola,
1985). It is important to know which of the feed components will
appear in the distillate stream from the extractive column in order
to design the equipment. Isovolatility curve is a reliable way, plot the
isovolatility curve on the composition triangle. If the curve inter-
sects the A-S edge, then component A will be the distillate from the
extractive column, otherwise component B will appear in the
distillate stream (Doherty and Malone, 2001; Hsu et al., 2010). A
relative volatility at various feed ratios plot is a useful tool to
demonstrate the effectiveness of the entrainer. The feed ratio is a
very important design variable. In fact, there is a minimum value of
feed ratio where the minimum and maximum reflux ratios occur
simultaneously. Below this value, we are unable to get the desired
purity of the product. With the sensitivity analysis tool in Aspen
Plus, it can clearly show the feasible range of feed ratio and reflux
ratio. It can help us choose the appropriate feed ratio and reflux
ratio to meet the product specifications.
3.2. Economical optimization
Once a base-case design is developed, the next step is to
optimize the design variables. It is conventional to use the total
annual cost (TAC) as the objective function to be minimized to
screen process candidates, which including annualized capital
costs and operating costs (Douglas, 1988; Muńoz et al., 2006;
Kraemer et al., 2009; Harwardt and Marquardt, 2012). The follow-
ing objective function was used to carry out a global economic
optimization.
 
TAC 103 $=year ¼ C v þ ir  FCI ð1Þ
Table 1
Utility prices.
Utility Price ($/GJ)
High pressure steam (41 barg) 17.7
Medium pressure steam (10 barg) 14.19
Low pressure steam (5 barg) 13.28
Cooling water 0.354
169
where FCI is the fixed capital investment. In this paper, just major
pieces of equipment were concerned, which are the three distilla-
tion column vessels (including column internals), three reboilers,
and four condensers. The “tray sizing” function in Aspen Plus was
used to calculate the diameter of column vessels and sieve plate is
selected in all columns. An overall heat transfer coefficient and a
differential temperature driving force in each heat exchange were
used to calculate the required heat transfer area. Here the overall
heat transfer coefficient and differential temperature for the
condenser and reboiler were all taken from Luyben (2006). A cost
estimation program CAPCOST of Turton was used to estimate all
the major equipment costs (Turton et al., 2009). ir is the fixed
capital recovery rate, it is assumed to be 0.3. Cv is the operating
cost, which includes mostly utilities consumption (steam, cooling
water). The utilities consumption costs can be calculated from the
heat duties of reboilers and condensers. High pressure steam with
pressure of 41 barg (254 1C), medium pressure steam with pres-
sure of 10 barg (184 1C) and low pressure steam with pressure of
5 barg (160 1C) and cooling water with temperature of 30 1C are
supposed to be available in the plant. The utility prices taken from
CAPCOST are listed in Table 1 (Turton et al., 2009).
The design variables need to be optimized in the conventional
process include the total stages of the PDC (NT1), the EDC (NT2),
and the ERC (NT3), fresh stream feeding location of the PDC (NF1),
entrainer and fresh streams feeding locations (NFE and NF2) of the
EDC, and the feeding location of the ERC (NF3) and the entrainer
flow rate (S). As so many design variables need to be optimized, a
calculation sequence was established to facilitate the optimization.
Such sequential iterative optimization procedure is clearly demon-
strated in Fig. 3 (Liang et al., 2014).
4. Case study: N-propanol and water separation
N-propanol is widely used as solvent or as important fine
chemical in the chemical and pharmaceutical production process.
It is impossible to separate n-propanol from water through
conventional distillation because of the existence of a binary
azeotrope. The n-propanol/water mixture has a minimum azeo-
trope with composition of about 42.5 mol% n-propanol at 1 atm.
Traditional separation method for the system include heteroge-
neous azeotropic distillation using cyclohexane or diisopropyl
ether (DIPE) as entrainer (Lee and Shen, 2003; Pienaar et al.,
2013). There is little report about separating this system through
extractive distillation. In this section, conventional extractive
distillation is studied for the separation of this system. The new
combined preconcentration/extractive distillation column setup is
applied to investigate the energy-saving possibility of this extrac-
tive distillation system.
4.1. Conventional three-column process
4.1.1. Steady state design analysis
The thermal condition specifications of the feed, products, and
recycle solvent streams are listed in Table 2. In all the alternatives
proposed the amount of product is always the same. The RADFRAC
distillation routine was used to design and simulate the processes.
The operating pressure was set at 1 atm for each column at both
the conventional and the new process. The NRTL thermodynamic
model was selected to predict the physical properties in the
simulation and built-in binary interaction parameters was used.
Table 3 shows the binary interaction parameters for the NRTL
model. The tray spacing for each column in the conventional and
the new process was specified as 0.4 m. In this work, we chose
N-methyl-2-pyrrolidone (NMP) as the heavy entrainer. From
Fig. 4a, it is observed that NMP is a feasible entrainer for the
170 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178

Fig. 3. Sequential iterative optimization procedure for the conventional extractive distillation system.

Table 2 automatically be assumed to be the distillate product removed

Specifications for streams.
from the EDC. From Fig. 4b, it is observed that NMP is an effective
Stream Conditions entrainer, causing an increase in the relative volatility of the water
and n-propanol, thus effectively “breaking” the azeotrope.
Feed Saturated liquid Beginning the design process from the PDC, the distillate
n-Propanol product Saturated liquid composition was specified to be 40.0 mol% n-propanol, a near-
H2O product Saturated liquid
NMP entrainer recycle 90.0 1C
azeotropic composition mixture of n-propanol-water. The bottom
composition was specified to be 99.9 mol% water to reduce the
n-propanol loss.
The EDC, it’s more complicated. The flow rate of the entrainer is
Table 3 an important design variable. Without appropriate entrainer flow
The binary interaction parameters for the Aspen plus NRTL model. rate and reflux ratio, we are unable to get the desired purity of the
product. With the sensitivity analysis tool in Aspen Plus, several
Binary parameters n-Propanol (i) n-Propanol (i) H2O(i)
case studies were made to investigate the effects of changing the
H2O(j) NMP(j) NMP(j)
flow rate of entrainer and reflux ratio on the product compositions
Aij 0.0 0.0 0.0 of the distillate stream from the EDC, which is the water product.
Aji 0.0 0.0 0.0 In each case, the number of theoretical plates (NT2), the distillate
Bij/K 181.2160  234.2208  120.5810 rate with different S and RR2 were fixed. Fig. 5a shows the effects
Bji/K 927.9748  205.8249 288.7615
of changing the flow rate of NMP and RR2 on the water product of
Cij 0.4687 0.3000 0.3055
the EDC with NT2 ¼58. Fig. 5a shows that at any value of the
entrainer flow rate, there is a minimum reflux ratio and also a
maximum reflux ratio, RRmin and RRmax get closer as we add less
separation of n-propanol-water system. That is because the entrainer. At small values of entrainer flow rate there will be a
entrainer NMP does not form any azeotropes with the original narrow range of RR2 over which the desired separation is possible.
components, and is completely miscible with them in all propor- For a given entrainer flow rate S, there is an optimum reflux ratio,
tions. The residue curve map has the optimal configuration-the which gives the desired water purity. It is critical to choose the
minimum boiling n-propanol-water azeotrope is the unstable feasible range of the entrainer flow rate and reflux ratio to meet
node, the entrainer NMP is the stable node, and the two compo- the water purity. To achieve the desired 99.9 mol% water purity,
nents of n-propanol and water are both saddles. From Fig. 4a, it is the entrainer flow rate must be above 290 kmol/h, and the reflux
observed that isovolatility curve intercepts the water-NMP edge of ratio RR2 is about 0.75. Below this value, the separation is not
the triangle. This observation determines that the overhead possible. Fig. 5b shows the effects of changing the flow rate of
product removed from the EDC should be water (the higher- NMP and RR2 on the impurity of n-propanol of the EDC.
boiling component) and not n-propanol (the lower-boiling com- Similar with the PDC, the ERC is a simple column. The bottom
ponent). It is worth mentioning that, in an extractive distillation composition of NMP purity was specified to be 99.999 mol%, the
system, the component with lower-boiling point cannot reason for the ultra-high purity in NMP is because the impurity in
 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178 171

H2O (100.02 C) Ternary Map (Mole Basis)

2.5
0.9

Relative Volatility
0.8
0.7 87.65 C 2.0
0.6
0.5
0.4 1.5
0.3
0.2
1.0
0.1 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
NMP
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
C3H8O-01 NMP/Azeotrope Feed Ratio
(203.99 C) (97.20 C)

Fig. 4. (a) Residue curve map with isovolatility curves for the N-propanol-H2O-NMP system at 1 atm. (b) Relative volatility at various feed ratios.

XD(mole frac of n-propanol)

1.000 0.005
XD(mole frac of H2O)

0.999 0.004 S=280kmol/h

S=310kmol/h
S=290kmol/h
0.998 0.003
S=300kmol/h S=300kmol/h
0.997 0.002
S=290kmol/h S=310kmol/h
0.996 S=280kmol/h 0.001

0.995 0
0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.8 1.0 1.2 1.4 1.6 1.8 2.0
RR2 RR2
Fig. 5. (a) Effect of RR2 and entrainer flow rate S in extractive column (NT2 ¼ 58) on H2O purity. (b) Effect of RR2 and entrainer flow rate S in extractive column (NT2 ¼ 58) on
impurity of n-propanol.

the NMP will have a very bad influence on the water purity. A heat 998.0
exchanger must be placed between the EDC and the ERC to cool 997.5
TAC(x1000$/year)

the NMP temperature. In this case, the temperature of the

entrainer NMP was cooled to be 90.0 1C. This is because Knight 997.0
and Doherty (1989) suggested that the entrainer feed temperature 996.5 NT1=9 with the
optimum NF1=6
to the EDC should be 5–15 1C below the boiling temperature of the 996.0
distillate product from the EDC.
995.5
4.1.2. Economical optimization 995.0
Fig. 6 shows the results of the optimization runs for the PDC 994.5
with NT1 and NF1 as the design variables. From this plot, the best 7 8 9 10 11
NT1 is stage 9 and the best NF1 is stage 6. NT1
Fig. 7a shows the summary of the TAC plots at each entrainer
Fig. 6. TAC vs NT1 with the optimum NF1 plot for the PDC.
flow rate for the EDC, it can be observed that the optimal entrainer
flow rate is 330 kmol/h. Fig. 7c shows the results for varying NT2
with the entrainer flow rate fixed at 330 kmol/h. From Fig. 7c, the variables of the conventional extractive distillation system can
optimal NT2 is 58. Fig. 7b shows the results of the optimization be directly applied to the new process. Thus, the original optimal
runs for varying NF2 and NFE with NT2 fixed at 58 and the entrainer configurations are all kept the same in the new process. The
flow rate fixed at 330 kmol/h. QR2 is chosen as the objective optimal design variables for the PDC and the EDC can be directly
function instead of using TAC to investigate the effect of changing applied to the preconcentration section and extractive section for
NF2 and NFE. From Fig. 7b, the optimal NFE is stage 6 and the the first column of the new process respectively. The final
optimal NF2 is stage 44. The final optimal design variables for the optimum process flow diagrams for the conventional extractive
EDC are S¼ 330 kmol/h, NT2 ¼ 58, NFE ¼ 6 and NF2 ¼ 44. distillation system and the new process are shown in Fig. 9.
Fig. 8 shows the summary of the TAC plots for the ERC. From
the figure, the optimal NT3 is stage 27 and the optimal NF3 is 4.3. Process comparison
stage 8.
In this section, heat duty, TAC of the above two processes will
4.2. New two-column process be compared. Table 4 shows the heat duty and the minimized TAC
results for the optimal conventional extractive distillation system
After the optimal design variables of the conventional extrac- and the new combined preconcentration/extractive column sys-
tive distillation system are determined, the optimal design tem. It is shown that the total reboiler duty of the new process is
172 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178

2372.0 3340
NFE=5
TAC(x1000$/year) 2371.5 NFE=6
3335
2371.0 NFE=7

QR2/kW
2370.5 3330

2370.0
3325
2369.5

2369.0 3320
320 330 340 350 43 44 45 46
S-solvent flow rate(kmol/h) NF2

TAC(x1000$/year) 2375

2374

2373

2372

2371

2370

2369
54 56 58 60
NT2
Fig. 7. (a) TAC plot at various entrainer flow rate S for the EDC. (b) QR2 vs NFE and NF2 for S¼ 330 kmol/h, NT2 ¼ 58 plots for the EDC. (c) TAC vs NT2 for S¼ 330 kmol/h plot for
the EDC.

1302 5. Case study: Ethyl acetate and ethanol separation

TAC(x1000$/year)

1301 Ethyl acetate (EA) is an important environmental friendly

solvent, and is industrially used in a wide range of applications.
1300 NT3=27 with the In the synthetic processes for ethyl acetate, ethyl acetate/ethanol
optimum NF3=8 mixture is commonly obtained. It is impossible to achieve the
1299 separation of ethyl acetate/ethanol mixture by conventional dis-
tillation, because the ethyl acetate/ethanol mixture forms a mini-
1298 mum azeotrope with composition of about 46.55 mol% ethanol at
1 atm. The purification techniques for ethyl acetate have been
1297 investigated, such as azeotropic distillation, extractive distillation,
25 26 27 28 29 30
and membrane separation (Berg, 1999; Zhang et al., 2008; Sato
NT3 et al., 2008). In this section, conventional extractive distillation is
Fig. 8. TAC vs NT3 with the optimum NF3 plot for the ERC. studied for the separation of this system. The new process is also
applied to investigate the energy-saving possibility of this extrac-
tive distillation system.

less than that of the conventional design by about 1575.9 kW,
which is about 22.76% saving in total reboiler energy consumption.
The reason is that part of the vapor flow required for the
preconcentration section is provided by the extractive section,
thereby causing the vapor flow provided by the intermediated
reboiler to be reduced. From the economical data, it is observed
that the separation using the new process leads to 15.63% reduc-
tion in total capital investment cost, that is because the new
process saves one condenser and the side stream reboiler is much
smaller compared to that of the conventional configuraion. There
is also a significant reduction in operating cost. The total operating
cost for the conventional process is 3371.4 (103 $/year) while the
new process costs 2752.4 (103 $/year). This represents a reduction
of 18.36% of the cost of the energy, that is mainly because the new
process leads to 22.76% saving in total reboiler energy consump-
tion. The TAC of the new process is 3918.0.0 (103 $/year), this can
be directly compared to the TAC of the conventional process. Thus,
the separation using the optimum new two-column desgin leads
to the reduction in total annual cost as much as 17.57%.
5.1. Conventional three-column process
5.1.1. Steady state design analysis
Table 5 shows the thermal condition specifications for the feed,
products, and solvent recycle stream. In all the alternatives
proposed the amount of product is always the same. The RADFRAC
distillation routine was used to design and simulate the processes.
The operating pressure was set at 1 atm for each column at both
the conventional and the new process. The NRTL thermodynamic
model was selected to predict the physical properties in the
simulation and built-in binary interaction parameters was used.
Table 6 shows the binary interaction parameters for the NRTL
model. The tray spacing for each column in the conventional and
the new process was specified as 0.4 m. In this work, we chose
furfural as the heavy entrainer for its low price. From Fig. 10a, it is
observed that furfural is a feasible entrainer for the separation of
ethyl acetate/ethanol mixture. From Fig. 10a, it is observed that
ethanol (the higher-boiling component) is the distillate of the EDC.
From Fig. 10b, it is observed that furfural is an effective entrainer,
 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178 173

Fig. 9. (a) The optimum process flow diagram for the new combined preconcentration/ extractive column system. (b) The optimum process flow diagram for the
conventional extractive distillation system.

Table 4 Table 5
Comparision between the optimum conventional design and corresponding new Specifications for streams.
two-column design.
Stream Conditions
Conventional process New process (%) Difference
Feed Saturated liquid
Heat duty Ethyl acetate product Saturated liquid
QRT (kW) 6925.3 5349.4  22.76 Ethanol product Saturated liquid
QCT(kW) 6876.9 5301.0  22.92 Furfural entrainer recycle 70.0 1C
Economical data
TIC (103 $) 4605.3 3885.3  15.63
TOC (103 $/year) 3371.4 2752.4  18.36 a near-azeotropic composition mixture of ethyl acetate/ethanol.
TAC (103 $/year) 4753.0 3918.0  17.57
For column 2, Fig. 11a shows the effects of changing the flow rate
of furfural and RR2 on the ethanol product of the EDC with
causing an increase in the relative volatility of the ethyl acetate NT2 ¼46. To achieve the desired 99.5 mol% ethanol purity, the
and ethanol, thus effectively “breaking” the azeotrope. furfural flow rate must be set above 390 kmol/h at a reflux ratio of
At the beginning of the design process from the PDC, the about 1.25. The feed ratio is about 2.36, which is much larger than
distillate composition was specified to be 45.0 mol% ethyl acetate, the feed ratio 1.94 in case 1 indicating that furfural to the ethyl
174 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178

acetate/ethanol system is less effective than NMP to the n-
propanol/water system. Fig. 11b shows the effects of changing
the flow rate of furfural and RR2 on the impurity of ethyl acetate
of the EDC. The bottom composition of the ERC was set at
99.999 mol% furfural. A heat exchanger must be placed between
the ERC and the EDC. In this case, the temperature of the entrainer
furfural was cooled to be 70.0 1C.
5.1.2. Economical optimization
Fig. 12 shows the results of the optimization runs for the PDC.
From this plot, the best NT1 is stage 27 and the best NF1 is stage 11.
The result shows that the separation of ethyl acetate/ethanol
system is much more difficult than the separation of n-propanol/
water system.
Fig. 13a shows the summary of the TAC plots at each entrainer
flow rate for the EDC, it can be observed that the optimal entrainer
flow rate is 540 kmol/h. The result also indicates that the entrainer
furfural to the ethyl acetate-ethanol system is less effective than
NMP to the n-propanol-water system. Fig. 13c shows the results
for varying NT2 with the entrainer flow rate fixed at 540 kmol/h.
From Fig. 13c, the optimal NT2 is 46. Fig. 13b shows the results of
the optimization runs for varying NF2 and NFE with NT2 fixed at 46
and the entrainer flow rate fixed at 540 kmol/h. From Fig. 13b, the
optimal NFE is stage 8 and the optimal NF2 is stage 31. The final
optimal design variables for the EDC are S ¼540 kmol/h, NT2 ¼46,
NFE ¼8 and NF2 ¼ 31.
Fig. 14 shows the summary of the TAC plots for the ERC. From
Fig. 14, the optimal NT3 is stage 25 and the optimal NF3 is stage 11.
section for the first column of the new process respectively. The
final optimum process flow diagram for the conventional extrac-
tive distillation system design and the new process design is
shown in Fig. 15.
5.3. Process comparison
In this section, heat duty, TAC of the above two processes for
the separation of ethyl acetate/ethanol were compared. Table 7
shows the heat duty and the minimized TAC results for the
optimal conventional extractive distillation system and the new
combined preconcentration/ extractive column system. It is shown
that the total reboiler duty of the new process is less than that of
the conventional design by about 1495.9 kW, which is about
17.25% saving in total reboiler energy consumption. From the
economical data, it is observed that the separation using the
new process leads to 14.10% reduction in total capital investment
cost. There is also a significant reduction in operating cost. The
total operating cost for the conventional process is 3748.7 (103 $/
year) while the new process cost is 3163.2 (103 $/year). This
represents a reduction of 15.62% of the cost of the energy, that is
mainly because the new process leads to 17.25% saving in total
reboiler energy consumption. The TAC of the new process is 4567.2
(103 $/year), this can be directly compared with the TAC of the
conventional process. Thus, the separation using the optimum
new two-column desgin leads to the reduction in total annual cost
by as much as 15.16%.

5.2. New two-column process 6. Summary and conclusion

Similar to case 1, after the optimal design variables of the In the extractive distillation process, the adding of high-boiling
conventional extractive distillation system are determined, the entrainer (organic solvent, ionic liquids) may reverse the relative
optimal design variables of the conventional extractive distillation volatility of the feed components causing the component with
system can be directly applied to the new process. Thus, the higher-boiling point be removed as the distillate from the extrac-
original optimal configurations are all kept the same in the new tive column. The energy-saving possibility of extractive distillation
process. The optimal design variables for the PDC and the EDC can system was investigated when this phenomenon occurred. Based
be directly applied to the preconcentration section and extractive on a three-column conventional extractive distillation system, an
innovative process by combining preconcentration column and
Table 6 extractive distillation column was proposed. The new process is
The binary interaction parameters for the Aspen plus NRTL model. energy efficient with lower capital investment. The main reason is
that part of the vapor flow required for the preconcentration
Binary parameters Ethanol (i) Ethanol (i) Ethyl acetate (i)
section is provided by the extractive section, thereby causing the
Ethyl acetate (j) Furfural (j) Furfural (j)
vapor flow provided by the intermediated reboiler to be reduced.
Aij 0.0 0.0 0.0 The new process is not limited to the studied systems which can
Aji 0.0 0.0 0.0 also be used to separate other azeotropic and in systems with close
Bij/K 216.3048 139.6013 548.0231 boiling point as long as the bottom product removed from the
Bji/K 95.0457 425.8981  257.2865
preconcentration column and the distillate product removed from
Cij 0.3000 0.30000 0.30000
the extractive distillation column have the same composition

ETHANOL (78.31 C) Ternary Map (Mole Basis)

2.8
0.9
2.6
Relative Volatility

0.8
2.4
0.7 2.2
0.6 2.0
72.14 C
0.5 1.8
0.4 1.6
0.3 1.4
0.2 1.2
1.0
0.1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
FURFURAL C4H8O-01 Furfural/Azeptrope Feed Ratio
(161.35 C) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 (77.20 C)

Fig. 10. (a) Residue curve map with isovolatility curves for the ethanol-ethyl acetate-furfural system at 1 atm. (b) Relative volatility at various feed ratios.
 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178 175

0.998 0.010

XD(mole frac of Ethanol)

0.997

XD(mole frac of EA)

0.009
0.996 0.008 S=380kmol/h
0.995 S=410kmol/h 0.007
S=390kmol/h
0.994 S=400kmol/h 0.006
0.993 0.005 S=400kmol/h
S=390kmol/h
0.992 0.004 S=410kmol/h
S=380kmol/h
0.991 0.003
0.990 0.002
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
RR2 RR2
Fig. 11. (a) Effect of RR2 and entrainer flow rate S in extractive column (NT2 ¼ 46) on ethanol purity. (b) Effect of RR2 and entrainer flow rate S in extractive column (NT2 ¼46)
on impurity of ethyl acetate.

1911 1457
TAC(x1000$/year)

TAC(x1000$/year)
1910
1456
1909 NT3=25 with the
NT1=27 with the
optimum NF3=11
optimum NF1=11
1908
1455
1907

1906 1454
25 26 27 28 29 30 31 24 25 26 27
NT1 NT3
Fig. 12. TAC vs NT1 with the optimum NF1 plot for the PDC. Fig. 14. TAC vs NT3 with the optimum NF3 plot for the ERC.

1927.0 3250
NFE=7
TAC(x1000$/year)

1926.5 3248 NFE=8

NFE=9
3246
QR2/kW

1926.0
3244
1925.5
3242
1925.0 3240

1924.5 3238
525 530 535 540 545 550 555 30 31 32 33
S-solvent flow rate(kmol/h) NF2

1932
TAC(x1000$/year)

1930

1928

1926

1924
42 44 46 48
NT2
Fig. 13. (a)TAC plot at various entrainer flow rate S for the EDC. (b) QR2 vs NFE and NF2 for S¼ 540 kmol/h, NT2 ¼46 plots for the EDC. (c) TAC vs NT2 for S¼ 540 kmol/h plot for
the EDC.
176 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178

F
300.000 kmol/hr 2 1atm 2 1atm
0.750 Ethanol 3395.5kW 1340.9kW
0.250 EA
11
225.758kmol/hr RR2=1.012
RR1=1.068
1908.4 kW 0.995Ethanol
0.00454 EA
0.000460Furfural D2
S MAKEUP 26
0.100 kmol/hr 11 74.347 kmol/hr
1.000 Furfural D1 0.00499 Ethanol
33 165.169kmol/hr 0.995 EA
0.550Ethanol 0.00001 Furfural
0.450 EA
2415.6 kW 56 5.0696E-21Furfural

70 B2 24
B1 536.721kmol/hr
611.068kmol/hr 1.2334E-16 Ethanol
0.000607Ethanol 1E-05 EA
3239.9kW 0.121 EA 0.99999 Furfural 2028.3 kW
0.878Furfural

Combined preconcentration/extractive Entrainer recovery

column column

1atm S 2
2 2 1atm 1atm
3391.8 kW MAKEUP 1481.4kW 1344.6kW
8
0.1kmol/hr
F 1.000 Furfural RR2=0.507 RR3=1.018
RR1=1.066
300.000 kmol/hr D2 D3
0.750 Ethanol D1 90.926 kmol/hr 74.342 kmol/hr
0.250 EA 11
11 165.169 kmol/hr 0.995 Ethanol 0.00499 Ethanol
0.550 Ethanol 0.004 EA 0.995 EA
0.450 EA 0.001 Furfural 0.00001 Furfural
31

26 45 B2 24 B3
B1 614.248 kmol/hr 539.906kmol/hr
134.831kmol/hr 0.000604 Ethanol 1.226E-16 Ethanol 2429.9kW
0.995 Ethanol 0.120 EA 1E-05 EA
3399.6kW 0.005 EA 3239.2 kW 0.879 Furfural 2033.7kW 0.99999 Furfural

Preconcentration Extractive distillation Entrainer recovery

column column column

Fig. 15. (a) The optimum process flow diagram for the new combined preconcentration/ extractive column system. (b) The optimum process flow diagram for the
conventional extractive distillation system.

specification. Two important systems were used to verify the
energy and economic advantages: the separation of n-propanol/
water and the separation of ethyl acetate/ethanol. Using these two
systems as examples, important observations of this paper are
summarized as below:
1. The separation of n-propanol/water using N-methyl-2-pyrroli-
done (NMP) as entrainer and the separation of ethyl acetate/
ethanol using furfural as entrainer causing the component with
higher-boiling point be removed as the distillate from the
extractive column, which are water and ethanol respectively
in these two case studies.
2. The new process is a useful technology to save energy and
lower the capital investment. The optimal design of the
conventional extractive distillation process and the new pro-
cess were presented. For the two case studies-the separation of
n-propanol/water and ethyl acetate/ethanol, the results show
that the new process leads to energy savings of 22.76% and
17.25% respectively and a similar reduction in capital invest-
ment and the total annual costs. In the new process, an
intermediate reboiler was used to meet the energy require-
ment in the preconcentration section.
Conflict of interest
The authors declare no competing financial interest.
 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178 177

Table 7 Errico, M., Pirellas, P., Torres-Ortega, C.E., Rong, B.-G., Segovia-Hernandez, J.G.,
Comparision between the optimum conventional design and corresponding new 2014a. A combined method for the design and optimization of intensified
two-column design. distillation systems. Chem. Eng. Process. 85, 69–76.
Errico, M., Rong, B.-G., Torres-Ortega, C.E., Segovia-Hernandez, J.G., 2014b. The
Conventional process New process (%) Difference importance of the sequential synthesis methodology in the optimal distillation
sequences design. Comput. Chem. Eng. 62, 1–9.
Muńoz, R., Montón, J.B., Burguet, M.C., Torre, J., 2006. Separation of isobutyl alcohol
Heat duty
and isobutyl acetate by extractive distillation and pressure-swing distillation:
QRT (kW) 8672.5 7176.6  17.25
simulation and optimization. Sep. Purif. Technol. 50, 175–183.
QCT(kW) 8647.8 7151.9  17.30
Luyben, W.L., 2006. Distillation Design and Control Using AspenTM Simulation, first
Economical data ed. John Wiley & Sons, New Jersey.
TIC (103 $) 5448.0 4680.0  14.10 Gil, I.D., Boti, D.C., Ortiz, P., Sanchez, O.F., 2009. Extractive distillation of acetone/
TOC (103 $/year) 3748.7 3163.2  15.62 methanol mixture using water as entrainer. Ind. Eng. Chem. Res. 48,
TAC (103 $/year) 5383.1 4567.2  15.16 4858–4865.
Gil, I.D., Gomez, J.M., Rodríguez, G., 2012. Control of an extractive ́distillation
process to dehydrate ethanol using glycerol as entrainer. Comput. Chem. Eng.
39, 129–142.
Nomenclature Harwardt, A., Marquardt, W., 2012. Heat-integrated distillation columns: vapor
recompression or internal heat integration? AIChE J. 58, 3740–3750.
Hsu, K.Y., Hsiao, Y.C., Chien, I.L., 2010. Design and control of dimethyl carbonate-
NRTL non-random two liquid methanol separation via extractive distillation in the dimethyl carbonate
E-DWC extractive dividing-wall column reactive-distillation process. Ind. Eng. Chem. Res. 49, 735–749.
Kiss, A.A., Suszwalak, D., 2012. Enhanced bioethanol dehydration by extractive and
PDC preconcentration distillation column azeotropic distillation in dividing-wall columns. Sep. Purif. Technol. 86, 70–78.
EDC extractive distillation column Knapp, J.P., Doherty, M.F., 1990. Thermal integration of homogeneous azeotropic
ERC entrainer recovery column distillation sequences. AIChE J 36, 969–984.
Knight, J.R., Doherty, M.F., 1989. Optimal design and synthesis of homogeneous
NF1 feeding location for the fresh feed
azeotropic distillation sequences. Ind. Eng. Chem. Res. 28, 564–572.
NFE feeding location for the entrainer Kossack, S., Kramer, K., Gani, R., Marquardt, W., 2008. A systematic synthesis
NF2 feeding location for the feed to column EDC framework for extractive distillation processes. Chem. Eng. Res. Des. 86,
NF3 feeding location for the feed to column ERC 781–792.
Kotai, B., Lang, P., Modla, G., 2007. Batch extractive distillation as a hybrid process:
NTn number of theoretical plates for column n separation of minimum boiling azeotropes. Chem. Eng. Sci. 62, 6816–6826.
QCn condenser duty for column n (kW) Kraemer, K., Kossack, S., Marquardt, W., 2009. Efficient optimization-based design
QRn reboiler duty for column n (kW) of distillation processes for homogeneous azeotropic mixtures. Ind. Eng. Chem.
Res. 48, 6749–6764.
QCT total condenser heat duty (kW) Langston, P., Hilal, N., Shingfield, S., Webb, S., 2005. Simulation and optimization of
QRT total reboiler heat duty (kW) extractive distillation with water as solvent. Chem. Eng. Proc 44, 345–351.
NMP N-methyl-2-pyrrolidone Lee, L., Shen, H., 2003. Azeotropic behavior of a waterþ n-propanolþ cyclohexane
mixture using cyclohexane as an entrainer for separating the water þn-
EA ethyl acetate proponal mixture at 760 mmHg. Ind. Eng. Chem. Res. 42, 5905–5914.
RRn reflux ratio for column n Lei, Z.G., Li, C.Y., Chen, B.H., 2003. Extractive distillation: a review. Sep. Purif. Rev.
TAC total annual costs 32, 121–213.
Lei, Z.G., Dai, C.N., Zhu, J.Q., Chen, B.H., 2014a. Extractive distillation with ionic
TIC total investment costs liquids: a review. AIChE J 60, 3312–3329.
TOC total operating costs Lei, Z.G., Xi, X.M., Dai, C.N., Zhu, J.Q., Chen, B.H., 2014b. Extractive distillation with
the mixture of ionic liquids and solid inorganic salt as entrainers. AIChE J 60,
2994–3004.
Liang, K., Li, W., Luo, H., Xia, M., Xu, C., 2014. Energy-efficient extractive distillation
process by combining preconcentration column and entrainer recovery column.
Acknowledgment Ind. Eng. Chem. Res. 53, 7121–7131.
Lladosa, E., Monton, J.B., Cruz Burguet, M., 2011. Separation of di-n-propyl ether and
n-propyl alcohol by extractive distillation and pressure swing distillation:
The authors thank for the support by the Programme of computer simulation and economic optimization. Chem. Eng. Process. 50,
Introducing Talents of Discipline to Universities (No: B06006) 1266–1274.
and the assistance from the staffs in the State Key Laboratories Luyben, W.L., 2008a. Effect of solvent on controllability in extractive distillation.
Ind. Eng. Chem. Res. 47, 4425–4439.
of Chemical Engineering (Tianjin University). Luyben, W.L., 2008b. Comparison of extractive distillation and pressure-swing
distillation for acetone methanol separation. Ind. Eng. Chem. Res. 47,
2696–2707.
References Modla, G., Lang, P., 2012. Removal and recovery of organic solvents from aqueous
waste mixtures by extractive and pressure swing distillation. Ind. Eng. Chem.
Abushwireb, F., Elakrami, H., Emtir, M., 2007. Recovery of aromatics from pyrolysis Res. 51, 11473–11481.
gasoline by conventional and energy-integrated extractive distillation. Comput. Modla, G., Lang, P., 2011. Comparison of extractive and pressure-swing batch
Aided. Chem. Eng. 24, 1071–1076. distillation for acetone-methanol separation. Comput. Aided. Chem. Eng. 29,
Arifin, S., Chien, I., 2008. Design and control of an isopropyl alcohol dehydration 382–386.
process via extractive distillation using dimethyl dulfoxide as an entrainer. Ind. Pacheco-Basulto, J.A., Hernández-McConville, D., Barroso-Muñoz, F.O., Hernández,
Eng. Chem. Res. 47, 790–803. S., Segovia-Hernández, J.G., Castro-Montoya, A.J., Bonilla-Petriciolet, A., 2012.
Berg, L., 1999. Separation of Ethyl Acetate from Ethanol by Azeotropic Distillation, Purification of bioethanol using extractive batch distillation: simulation and
US Patent 5993610. experimental studies. Chem. Eng. Process. 61, 30–35.
Doherty, M.F., Caldarola, G.A., 1985. Design and synthesis of homogeneous Pienaar, C., Schiwarz, C.E., Knoetze, J.H., Burger, A.J., 2013. Vapor–liquid–liquid
azeotropic distillations. 3. The sequencing of columns for azeotropic and equilibria measurements for the dehydration of ethanol, isopropanol, and
extractive distillations. Ind. Eng. Chem. Fundam 24, 474–485. n-propanol via azeotropic distillation using DIPE and isooctane as entrainers.
Doherty, M.F., Malone, M.F., 2001. Conceptual Design of Distillation Systems, first J. Chem. Eng. Data. 58, 537–550.
ed. McGraw Hill, New York. Sato, K., Sugimoto, K., Nakane, T., 2008. Separation of ethanol/ethyl acetate mixture
Douglas, J.M., 1988. Conceptual Design of Chemical Processes. McGraw-Hill, by pervaporation at 100  130 1C through NaY zeolite membrane for industrial
New York. purpose. Micropor. Mesopor. Mater 115, 170–175.
Errico, M., Rong, B.-G., Tola, G., Spano, M., 2013a. Optimal synthesis of distillation Seader, J.D., Henley, E.J., 1998. Separation Process Principles, first ed. John Wiley &
systems for bioethanol separation. Part 1: Extractive distillation with simple Sons, New Jersey.
columns. Ind. Eng. Chem. Res. 52, 1612–1619. Sucksmith, I., 1982. Extractive distillation saves energy. Chem. Eng. 28, 91–95.
Errico, M., Rong, B.-G., Tola, G., Spano, M., 2013b. Optimal synthesis of distillation Turton, R., Bailie, R.C., Whiting, W.B., Shaeiwitz, J.A., 2009. Analysis, Synthesis and
systems for bioethanol separation. Part 2: Extractive distillation with complex Design of Chemical Processes, third ed. Prentice Hall, New Jersey.
columns. Ind. Eng. Chem. Res. 52, 1620–1626. Wang, Q., Yu, B., Xu, C., 2012. Design and control of distillation system for methylal/
Errico, M., Rong, B.-G., Tola, G., 2013c. Optimal synthesis and design of extractive methanol separation: Part 1: Extractive distillation using DMF as an entrainer.
distillation systems for bioethanol separation: from simple to complex col- Ind. Eng. Chem. Res. 51, 1281–1292.
umns. Comput. Aided. Chem. Eng. 32, 373–378. Widagdo, S., Seider, W.D., 1996. Azeotropic distillation. AIChE J 42, 96–130.
178 Y. An et al. / Chemical Engineering Science 135 (2015) 166–178

Wu, Y.C., Hsu, P.H.C., Chien, I.L., 2013. Critical assessment of the energy-saving
potential of an extractive dividing-wall column. Ind. Eng. Chem. Res. 52,
5384–5399.
Xia, M., Yu, B., Wang, Q., Jiao, H., Xu, C., 2012. Design and control of extractive
dividing-wall column for separating methylal-methanol mixture. Ind. Eng.
Chem. Res. 51, 16016–16033.
Xu, S.L., Wang, H.Y., 2006. A new entrainer for separation of tetrahydrofuran–water
azeotropic mixture by extractive distillation. Chem. Eng. Process. 45, 954–958.
Zhang, D.L., Deng, Y.F., Li, C.B., Chen, J., 2008. Separation of ethyl acetate-ethanol
azeotropic mixture using hydrophilic ionic liquids. Ind. Eng. Chem. Res. 47,
1995–2001.