Available online at www.sciencedirect.

com

Journal of Process Control 18 (2008) 215231

www.elsevier.com/locate/jprocont

Overall control strategy of a coupled reactor/columns

process for the production of ethyl acrylate
I-Lung Chien *, Kay Chen, Chien-Lin Kuo
Department of Chemical Engineering, National Taiwan University of Science and Technology, Taipei 106, Taiwan
Received 1 December 2006; received in revised form 10 February 2007; accepted 20 February 2007

Abstract
Ethyl acrylate (EA) is widely used in industry as a precursor for varnishes, adhesive, and nishes of papers and textiles. This important ester can be produced directly from ethanol (EtOH) and acrylic acid (AA) via esterication reaction with the presence of sulphuric
acid as homogeneous catalyst. The proposed design owsheet of this process includes a CSTR reactor coupled with a rectier and an
overhead decanter. In order to further purify the nal EA product, another stripper is needed with its top vapor recycled back to decanter. The simplest and industrial easily applicable overall control strategy will be investigated with only one tray temperature control loop
in each of the two columns. The nal proposed overall control strategy of this process is found to be dierent than another similar coupled reactor/columns process published earlier [I-L. Chien, Y.P. Teng, H.P. Huang, Y.T. Tang, Design and control of an ethyl acetate
process: coupled reactor/column conguration, J. Proc. Cont. 15 (2005) 435449]. Both EtOH and AA feed ow rates are used as manipulated variables in the overall control strategy with CSTR heat duty left as throughput manipulator for the overall process. The nal EA
product with stringent specications of 0.1 wt% EtOH and 0.005 wt% AA impurities can be achieved with this proposed overall control
strategy despite feed ow rate and feed composition disturbances.
 2007 Elsevier Ltd. All rights reserved.
Keywords: Ethyl acrylate; Esterication reaction; Reactive distillation; Coupled reactor/columns; Optimum design; Overall control strategy

1. Introduction
Ethyl acrylate (EA) is widely used in industry as a precursor for varnishes, adhesive, and nishes of papers and
textiles. This important ester can be produced directly from
ethanol (EtOH) and acrylic acid (AA) via esterication
reaction with the presence of sulphuric acid as homogeneous catalyst. Only in a very recent paper [2], the kinetics
of this esterication reaction has been given. There is no
paper in the literature on the subject of the production of
ethyl acrylate, thus the results of this paper should be useful to other researchers.
In this study, overall control strategy of this process
with coupled reactor/columns conguration will be stud-

Corresponding author. Tel.: +886 2 27376652; fax: +886 2 27376644.

E-mail address: chien@ch.ntust.edu.tw (I-Lung Chien).

0959-1524/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jprocont.2007.02.006

ied. The principal behind the coupled reactor/columns conguration, similar to reactive distillation, is that the
continuous removal of products from the esterication
reaction mixture by distillation reduces the backward reaction rate. The advantage of the coupled reactor/columns
conguration over reactive distillation according to Yi
and Luyben [3] include: the existing reactor/columns in
the plant can be retrotted for this usage; easy maintenance
of the overall system; larger reactor holdup and dierent
reaction temperature can easily be designed; etc.
In a three-paper series by Yi and Luyben [35], they
studied the design and control of various coupled reactor/column systems. The studied systems include: a binary
reactor/rectier, a binary reactor/stripper, a multicomponent reactor/rectier, a multicomponent reactor/rectier/
stripper, and a more complex process that consists of a
coupled reactor/stripper, two distillation columns and
one recycle stream. Their studied systems are very simple,

216

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

ideal chemical systems and also no liquidliquid equilibrium is considered. Chiang et al. [6] studied a coupled reactor/column system for the production of amyl acetate.
Their process is much simpler than the studied ethyl acrylate process because amyl acetate has the highest boiling
point in the system. Their system with the conguration
of reactor with rectier on top and stripper on the bottom
produces amyl acetate from the bottom of the stripper and
also produces water through aqueous phase of a decanter.
Chien et al. [1] proposed a coupled reactor/column conguration for the production of ethyl acetate. Their process
owsheet conguration is very similar to the ones in this
paper. However, the overall control strategy is dierent
than the one will be developed for this system.
The organization of this paper is as follows. The thermodynamic properties of this four-component system and
the kinetics of this esterication reaction will be given in
Section 2. The design owsheet of a complete coupled reactor/column system will be proposed in Section 3. A nal
EA product purity of over 99.5 wt% will be obtained in
the proposed design with stringent specications of
0.1 wt% EtOH and 0.005 wt% AA impurities. In Section
4, the overall control strategy of this process will be investigated. Only one tray temperature control loop in each of
the two columns (rectier and stripper) will be used. Feed
ow rate and feed composition disturbances will be used
to test the overall control strategy. Some concluding
remarks will be drawn in the nal section.

2. Thermodynamic and kinetic model used in the simulation

There are total of four azeotropes in this system including two homogeneous azeotropes of EtOH + H2O and
EtOH + EA and two heterogeneous azeotropes of
EA + H2O and EtOH + EA + H2O. In order to accurately
represent the overall system, liquid activity coecient
model was used for the vaporliquidliquid equilibrium.
A suitable NRTL (nonrandom two-liquid) model parameter set has been established with excellent prediction of the
compositions and temperatures for the four azeotropes in
this system. In this NRTL parameter set, the Aspen Plus
built-in binary-pair parameters of AAEA and AAH2O
were used. For the EtOHH2O, EtOHEA, EAH2O pairs,
binary parameters were obtained to t well the LLE
boundary of these three components. For the one pair of
AAEtOH that does not have the Aspen Plus built-in
NRTL binary parameters, the Dortmund modied UNIFAC group contribution estimation method [7,8] was used
to obtain the remaining thermodynamic model parameters.
Vapor association of Acrylic acid due to dimerization has
also been included by using the second virial coecient
of the HaydenOConnell [9] model in the vapor phase.
The Aspen Plus built-in association parameters were
employed to compute fugacity coecient.
The kinetic model of this esterication reaction is from
the paper by Witczak et al. [2] with diluted sulphuric acid

as homogeneous catalyst. The reaction can be seen as

below:
C2 H5 OH CH2 CHCOOH
AA

EtOH

$ CH2 CHCOOC2 H5 H2 O

EA

The kinetic equation is

r k1 C cat C 2AA C 2EtOH 

C 2EA C 2H2 O
K2

!
mol=dm3 min

where


15900
k1 3:26  10 exp
RT


6490
4
K 2:71  10 exp
RT
6

dm12 =mol4 min

and R is gas constant (1.987 cal/mol/K) and Ccat is assumed to have value of 0.15 mol/dm3. In the kinetic equation, all concentrations are with unit of (mol/dm3) and
temperature in K.
3. Design owsheet of the complete process
The RCM for the EtOHEAH2O three-component
system and AAEAH2O three-component system can be
seen in Figs. 1 and 2. From these two gures, the highest
boiling point temperature of the whole system including
the pure components and azeotropes is the acrylic acid
(AA) at 141.19 C and the lowest temperature of the whole
system is the EtOH + EA + H2O three-component azeotrope at 75.29 C. The two products (ethyl acrylate and
water) of this esterication reaction are neither the lightest
nor the heaviest component in the system, thus the complete designed process will need to be more complex in
comparison with the other reactive distillation papers in
the literature.
The proposed design of this process including a CSTR
reactor coupled with a rectier (without heat source). The
heat input in the CSTR totally vaporizes the reactor outlet
stream to vapor phase and continuously enter the rectier
from the bottoms to promote the forward reaction further.
The bottom liquid stream from the rectier containing
mostly heavy boiler AA is recycled back to the CSTR.
The composition of the top vapor stream from the rectier
is close to the lightest boiler of ternary azeotrope of EtOH + EA + H2O. This stream after sub-cooling to 40 C
can be naturally separated inside a decanter to form
organic and aqueous phases. Extra water is added in the
decanter to maintain suitable composition inside of the
liquidliquid boundary. The water purity of the aqueous
phase is quite high, thus it can go to a waste water treatment plant for discharge. The organic phase composition
by natural liquidliquid separation has the benet of crossing the distillation boundary into a desirable region to
obtain pure ethyl acylate product (see Fig. 1). This organic

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

217

EtOH
(78.31 C)

0.9

(78.20 C)

0.4

0.5

0.6

0.7

0.8

(77.87 C)

0.1

0.2

0.3

(75.29 C)

EA

WATER

(99.40 C)

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

(100.02 C)

(81.04 C)

Fig. 1. RCM of EtOHEAH2O three-component system.

AA

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

(141.19 C)

EA
(99.40 C)

WATER
0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

(100.02 C)

(81.04 C)

Fig. 2. RCM of AAEAH2O three-component system.

phase stream is partly reuxed and is partly designed to

feed into another stripper with reboiler for further purica-

tion into the nal EA product. The top vapor of this stripper with composition near the top vapor of the rectier is

218

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

Water

Decanter
Aqueous
Organic
Reflux

AA
EtOH
Steam
CSTR

Reboiler

Steam

EA

Fig. 3. Conceptual design of the overall process owsheet.

cost. The reason to add product value and the costs of

two feed streams in the TAP calculation is because the
two feeds are not necessarily xed at equal molar ratio.
The annualized capital costs follow directly from the calculation procedure in Douglas [10] with the annual capital
charge factor of 1/3 was used. The annualized equipment
cost includes CSTR, column shells, column trays, reboilers,
and condensers. The utility cost including the steam and
cooling water costs are calculated the same way as in
Chiang et al. [6]. The waste water treatment cost is calcu-

5.60e+5

5.58e+5

5.55e+5

5.56e+5

5.50e+5

Profit ($/year)

Profit ($/year)

also condensed and then fed into the decanter. The bottom
stream of the stripper is the nal EA product with stringent
specications of 0.1 wt% EtOH and 0.005 wt% AA impurities in this product stream. This conceptual design of the
overall process owsheet can be seen in Fig. 3.
The design owsheet is selected based on the maximization of Total Annual Prot (TAP) for the overall system.
This TAP includes: the product value minus the costs of
two feed streams, minus annualized capital costs, minus
total utility costs, and minus the waste water treatment

5.45e+5
5.40e+5
5.35e+5

Column 1=24 stages

Column 2=7 stages
CSTR volume=126 cum

5.30e+5
20

40

60

80

100

5.52e+5
5.50e+5
Water flow rate=50 mole/min
Column 1=24 stages
Column 2=7 stages

5.48e+5

5.46e+5
125

5.25e+5
0

5.54e+5

120

126

127

Water flow rate (mol/min)

129

130

131

132

133

5.568e+5

5.568e+5

5.567e+5

Profit ($/year)

5.567e+5

Profit ($/year)

128

CSTR Volume (cum)

5.566e+5
5.565e+5
5.564e+5
Water flow rate=50 mole/min
CSTR volume=126 cum
Column 2=7 stages

5.563e+5

5.566e+5
5.565e+5
5.564e+5
Water flow rate=50 mole/min
CSTR volume=126 cum
Column 1=24 stages

5.563e+5
5.562e+5
5.561e+5

5.562e+5
22

23

24

Tray of Column 1

25

26

Tray of Column 2

Fig. 4. Optimization result with EtOH feed ow rate of 96 mol/min.

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

lated with the estimation of $0.053/m3 (given in Table 3.4

of Turton et al. [11] textbook).
The design and operating variables that need to be
determined include AA/EtOH feed ratio, CSTR holdup,
total stages of the rectier, total stages of the stripper,
and the water addition rate into the decanter. An iterative
sequential optimization procedure is proposed to nd the
optimal owsheet of the overall system. In the design of
the process owsheet, pure AA feed composition is
assumed while the EtOH feed stream is practically assumed
to contain 82.2 mol% EtOH and 17.8 mol% H2O.
The acrylic acid ow rate is xed at 66 mol/min to set
the throughput of the overall process. The organic reux
ow rate from the decanter is manipulated to hold the
AA impurity in the nal EA product to be at the specication of 0.005 wt%. The reboiler duty of the stripper is
manipulated to hold the EtOH impurity in the nal EA
product to be at the specication of 0.1 wt%. The iterative
sequential optimization procedure is to nd the optimization result at a particular EtOH feed ow rate. In each
optimization search, four design and operating variables
can be changed including CSTR holdup volume, rectifying
column total stage, stripping column total stage, and water
addition rate. Iterative procedure is needed to nd the optimization result. Fig. 4 shows an example of the optimization result with EtOH feed ow rate at 96 mol/min. In
this case, the water addition rate to achieve the highest
TAP is at 50 mol/min, the CSTR holdup volume is
126 m3, rectifying column total stage is 24, and stripping
column total stage is 7. Notice that in this case, CSTR
holdup volume below 126 m3 cannot satisfy the AA impurity specication in nal EA product stream.
Summarizing the optimization results at various EtOH
feed ow rates, Fig. 5 shows that the optimal EtOH feed
ow rate is at 102 mol/min. At this ow rate, the TAP is
maximized at $5.95 105. The optimum pure AA and pure
EtOH feed ratio is calculated to be 1:1.27, not at exactly
equal molar ratio. With this optimum feed ratio, other
optimum design and operating variables are: water addi-

6.0e+5

Profit ($/year)

5.9e+5

5.8e+5

5.7e+5

5.6e+5

5.5e+5
94

96

98

100

102

104

106

108

110

EtOH flow rate (mol/min)

Fig. 5. Summary of optimization results at various EtOH feed ow rates.

219

Table 1
Optimum process design for this system
CSTR holdup (m3)

64

Rectier total stages

(where 20th stage is the CSTR)
Stripper total stages
(including the reboiler)
Fresh EtOH feed ow rate
(g mol/min)
EtOH feed composition

20

Fresh AA feed ow rate

(g mol/min)
AA feed composition
Vapor ow rate from CSTR
to rectier (g mol/min)
Bottom liquid ow rate from rectier to
CSTR (g mol/min)
CSTR heat duty (KW)
Water injection rate into the
decanter (g mol/min)
Organic reux ow rate (g mol/min)
Organic outlet ow rate into
stripper (g mol/min)
Stripper reboiler duty (KW)
Aqueous outlet ow rate (g mol/min)
EA product ow rate (g mol/min)
EA product composition

7
102
82.2 mol% EtOH
17.8 mol% H2O
66
100 mol% AA
258
90
151.8
100.0
84.05
145.55
70.10
204.1
63.9
99.78 mol% (99.89 wt%) EA
0.0069 mol% (0.005 wt%) AA
0.213 mol% (0.10 wt%) EtOH
1 104 mol% H2O

tion rate at 100 mol/min, CSTR holdup volume at 64 m3,

rectifying column total stage at 20, and stripping column
total stage at 7. The nal overall process owsheet can be
seen in Table 1. Notice from the nal owsheet, the specications of the two impurities are met and the nal EA
product purity is at 99.78 mol% (or 99.89 wt%).
4. Overall control strategy of this process
Some straightforward inventory and other loops are
determined rst. These include the following: the organic
phase level is controlled by the organic outlet ow to
stripper; the aqueous phase level is controlled by the aqueous outlet ow; the bottom level of the stripper is controlled by the nal product ow; the top pressures of
the rectier and stripper are controlled at 1.1 atm by the
top vapor ow; and the temperature at decanter are controlled at 40 C by the condenser duty. The organic reux
ratio is xed by a ratio scheme where the value of the
ratio can be set by a tray temperature control loop if necessary. The extra water ow rate into the decanter is ratio
to a throughput manipulator yet to be determined. Summarizing of the basic loops in the process can be seen in
Fig. 6.
For the candidate overall control strategies studied in
this paper, some other prerequisite assumptions are outlined below. Firstly, all control loops are in conventional
PID form for easier industrial applications. Secondly,

220

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

Water
FC

Ratio to a
throughput
manipulator

TC

TC

Decanter
LC

PC

Organic
Reflux

LC

Aqueous

PC

1
FC
X

FC

AA
FC

19

EtOH

6
LC

Steam
7

20

CSTR

Reboiler

Steam
EA

Fig. 6. Basic regulatory control strategy of the overall process.

only one tray temperature control loop is investigated in

each of the rectier and stripper column to avoid strong
loop interactions. Thirdly, the manipulated variable of
the tray temperature control loop for the stripper are
straightforwardly determined as the stripper reboiler
duty.
The manipulated variables of the remaining two most
important control loops are yet to be determined. These
two control loops are the CSTR level control loop and
the rectier tray temperature control loop. The CSTR level
control loop unlike the other level control loops needs to
include integral mode to make sure the reaction volume
is maintained. The candidate manipulated variables for
these two important loops are: AA feed ow, EtOH feed
ow, CSTR heat duty, and organic reux ratio. Thus, the
planning of the overall control strategy is to select two
out of the above four manipulated variables to control
CSTR level and one tray temperature at rectier. For the
remaining two manipulated variables, one can be used as
the throughput manipulator to set the production rate of
the overall process, and the other is xed throughout various disturbance changes.
4.1. Determine of tray temperature control point and the
control structure
Closed-loop sensitivity analysis similar to the one used
in Lee et al. [12] will be used here to determine the tray
temperature control point at rectier and also at stripper.
The purpose for the closed-loop sensitivity analysis is different in this paper. In Lee et al. [12], the control structure
was already set and the closed-loop analysis is solely used
to determine the temperature control point. However, in
this paper, the closed-loop sensitivity analysis is not only
used to determine the tray temperature control point at

rectier and at stripper, but it is also used to screen out

the possible worse control structures in the overall control
strategy.
Because there are four free manipulated variables
which could be used for the CSTR level control loop and
rectier tray temperature control loop, ve possible control
structures considered are listed below:
CS1: EtOH feed ow and CSTR heat duty are used as
manipulated variables for the CSTR level and rectier tray temperature loops, AA feed ow is used as
throughput manipulator, and organic reux ratio is
xed throughout various disturbance changes.
CS2: AA feed ow and CSTR heat duty are used as
manipulated variables for the CSTR level and rectier tray temperature loops, EtOH feed ow is used
as throughput manipulator, and organic reux ratio
is xed throughout various disturbance changes.
CS3: EtOH feed ow and AA feed ow are used as manipulated variables for the CSTR level and rectier tray
temperature loops, CSTR heat duty is used as
throughput manipulator, and organic reux ratio is
xed throughout various disturbance changes.
CS4: EtOH feed ow and organic reux ratio are used as
manipulated variables for the CSTR level and rectier tray temperature loops, AA feed ow is used as
throughput manipulator, and CSTR heat duty is
xed throughout various disturbance changes (only
ratio to measurable AA feed ow).
CS5: AA feed ow and organic reux ratio are used as
manipulated variables for the CSTR level and rectier tray temperature loops, EtOH feed ow is used
as throughput manipulator, and CSTR heat duty is
xed throughout various disturbance changes (only
ratio to measurable EtOH feed ow).

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

Organic reflux ratio fixed & AA fixed

1.0

Delta temperature (C)

0.5

0.0

-0.5

-1.0

-1.5
1

9 10 11 12 13 14 15 16 17 18 19 20

Trays of 1ST Column

0.4
AA-5%
EtOH(H2O)+20%
EtOH(H2O)-20%
throughput+10%
throughput-10%

0.3

Delta temperature (C)

There is another control structure which uses CSTR

heat duty and organic reux ratio as manipulated variables
for the CSTR level and rectier tray temperature loops.
However, in this case, both EtOH feed ow and AA feed
ow will be xed, thus will be infeasible to cope with feed
composition changes.
The screening of the above ve possible control structures can easily be made by the closed-loop sensitivity
analysis described below using process simulation tool.
In the following, two ideal composition control loops are
assumed to be present which holds the specications of
0.1 wt% EtOH and 0.005 wt% AA impurities in the product stream by varying two chosen manipulated variables.
One of the manipulated variables has been pre-determined
to be the stripper reboiler duty which controls the EtOH
impurity at 0.1 wt% in the product stream and the other
manipulated variable to hold AA impurity at 0.005 wt%
in the product stream can be varied due to dierent control structure. Three disturbance changes are made in
the closed-loop simulations, they are 10% changes in
the throughput manipulator; 20% changes in the EtOH
feed water concentration; and the changes of water concentration in the AA feed from 0 mol% to 5 mol%. For
example, the closed-loop simulation for control structure
CS1 can be achieved with: xing AA impurity in the product stream at 0.005 wt% by varying EtOH feed ow rate;
xing CSTR level at original setpoint by varying CSTR
heat duty; and xing EtOH impurity in the product stream
at 0.1 wt% by varying stripper reboiler duty. Thus, for
control structure CS1 in the simulation runs, organic
reux ratio and AA feed rate are xed with AA feed rate
only moves when throughput manipulator changes are
required.
For each disturbance case, the temperature proles of
rectier and stripper under perfect composition control
can be obtained for the above mentioned three disturbance
changes. By comparing the temperature proles for the
disturbance changes versus the base case, the deviations
of the temperature proles of rectier and stripper can
be plotted. Because this is an ideal disturbance rejection
condition with both product specications hold at their
original values, one could choose a tray temperature control point for rectier and another one for stripper and
using two tray temperature control loops to replace the
ideal dual-composition control loops. The chosen temperature control point for rectier or stripper should be the
one with the least deviation of the tray temperature.
Fig. 7 shows such plot for control structure CS1 under
above mentioned three kinds of disturbance changes.
The focus of observing the least deviations of the temperatures should be made for the feed composition changes.
The reason is because the disturbance change for the
throughput manipulator is considered as known disturbance, thus, some calculating scheme to adjust the setpoint
value of the temperature control loop as proposed in
Huang et al. [13] can easily be applied to hold the product
specications.

221

0.2

0.1

0.0

-0.1

-0.2
1

Trays of 2nd Column

Fig. 7. Closed-loop sensitivity plot with organic reux ratio and AA feed
xed.

From Fig. 7, the suitable choice of the tray temperature

control point for rectier is at tray #5 and for stripper is
also at tray #5. Notice that for rectier, the tray #5 gave
very small deviations under 20% changes in the EtOH
feed water concentration, however, the temperature deviation under 5% change in the AA feed concentration gave
considerably larger temperature deviation. From these
closed-loop simulations, one could predicts that by using
control structure CS1 with two tray temperature control
points at tray #5 of rectier and tray #5 at stripper, the
control performance should perform well under 20%
changes in the EtOH feed water concentration. However,
for 5% changes in the AA feed concentration there will
be some deviations in nal product impurity especially
for AA impurity. Of course the chosen tray temperature
control points should be evaluated so that enough openloop sensitivity is there between the manipulated variable
(in this case EtOH feed rate) and rectier tray #5 temperature and also between stripper reboiler duty and stripper

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

tray #5 temperature. This open-loop sensitivity evaluation

will be shown later in this section.
Similar closed-loop sensitivity simulations can be run
for other control structures CS2, CS3, CS4, and CS5. Figs.
811 show the temperature deviations for control structures CS2CS5, respectively. By observing these plots,
the tray temperatures for various control structures with
the least temperature deviation for rectier are at tray
#5 for CS2 and CS3 and at tray #4 for CS4 and CS5.
For the stripper, the one with the least temperature deviation are all at tray #5. The control at stripper should be
easy and should give acceptable control performance
because the temperature deviations at tray #5 under various disturbances are all not very large. On the other hand,
the control at rectier is more dicult and should draw
more attention. The control structure CS3 gave the least
temperature deviations at tray #5 of rectier under various disturbance changes while CS4 and CS5 gave quite
large temperature deviations at the least temperature deviation of tray #4 under 5% change in the AA feed
concentration.

Organic reflux ratio fixed & Q1 fixed

1.0
0.8
0.6

Delta temperature (C)

222

0.4
AA-5%
EtOH(H2O)+20%
EtOH(H2O)-20%
throughput+10%
throughput-10%

0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Trays of 1ST Column

0.20
0.15

Delta temperature (C)

0.10

Organic reflux ratio fixed & EtOH fixed

1.5

Delta temperature (C)

1.0

0.5

0.05
0.00
-0.05
-0.10

0.0

-0.15

-0.5

-0.20
1

Trays of 2nd Column

-1.0

Fig. 9. Closed-loop sensitivity plot with organic reux ratio and CSTR
duty xed.
-1.5
1

9 10 11 12 13 14 15 16 17 18 19 20

Trays of 1ST Column

0.4

AA-5%
EtOH(H2O)+20%
EtOH(H2O)-20%
throughput+10%
throughput-10%

0.3

Delta temperature (C)

From the above simply closed-loop sensitivity analysis,

we can abandon further dynamic investigation of control
structures CS4 and CS5 and focus on CS1 to CS3 rst.
Only if dynamic behaviors of CS1 to CS3 are not acceptable, we will re-investigate CS4 and CS5 then.

0.2

4.2. Loop pairing and dynamic considerations

0.1
0.0
-0.1
-0.2
-0.3
-0.4
1

Trays of 2nd Column

Fig. 8. Closed-loop sensitivity plot with organic reux ratio and EtOH
feed xed.

From the above analysis, it is better to x organic reux

ratio, thus, the manipulated variables used for the rectier
tray temperature at tray #5 and CSTR level loops should
be picked from the below three manipulated variables:
EtOH feed; AA feed; or CSTR heat duty. Fig. 12 shows
the open-loop dynamic response with organic reux ratio
xed and step changes of either EtOH feed, AA feed, or
CSTR duty. When one of the three manipulated variables
step changes, the remaining two manipulated variables
are kept at their original base case values. The rst observation is that none of the dynamic responses can be mod-

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

Q1 fixed & EtOH fixed

1.0

1.0

0.5

0.5

Delta temperature (C)

Delta temperature (C)

Q1 fixed & AA fixed

0.0

-0.5

-1.0

0.0

-0.5

-1.0

-1.5

-1.5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

1 2 3 4 5 6 7 8

Trays of 1ST Column

0.6

9 10 11 12 13 14 15 16 17 18 19 20

Trays of 1ST Column

0.6

AA-5%
EtOH(H2O)+20%
EtOH(H2O)-20%
throughput+10%
throughput-10%

0.2

0.0

-0.2

-0.4

AA-5%
EtOH(H2O)+20%
EtOH(H2O)-20%
throughput+10%
throughput-10%

0.4

Delta temperature (C)

0.4

Delta temperature (C)

223

0.2

0.0

-0.2

-0.4

-0.6

-0.6
1

Trays of 2nd Column

Fig. 10. Closed-loop sensitivity plot with CSTR duty and AA feed xed.

eled as rst-order plus deadtime process. A suitable model

to use should be the integrating plus deadtime model form.
Also from this gure, one observed that the dynamic
response of CSTR heat duty versus tray #5 temperature
of rectier is very problematic, thus this pairing should
be avoid. The inverse response behavior can be explained
by the combinatory eects of this tray temperature due
to CSTR heat duty and also from the changing of the
CSTR level (reactor volume). It is also notice that the
open-loop eect of AA feed to tray #5 temperature of rectier is considerably smaller than that from EtOH feed,
thus, it is better to use EtOH feed to control tray #5
temperature.
Since CSTR heat duty is not suitable to use as the
manipulated variable for the tray #5 temperature control
loop, one of the candidate overall control strategy is to
use CSTR heat duty to control CSTR level. This pairing
was also used in Chien et al. [1]. Fig. 13 shows another
reason why AA feed is not suitable to control tray #5
temperature. With CSTR heat duty manipulating CSTR

Trays of 2nd Column

Fig. 11. Closed-loop sensitivity plot with CSTR duty and EtOH feed
xed.

level in automatic mode, we can do open-loop step

response for changes in AA or EtOH feed ow rate.
From Fig. 13, one observe that the open-loop response
with AA feed as the manipulated variable exhibits large
inverse response. Let us take AA feed +1% changes as
an example. Because AA is a high-boiler with normal
boiling point at 118 C, thus increasing AA feed into
the system should eventually make tray #5 temperature
to increase. However, increasing AA feed ow rate will
also result in increasing the CSTR level, thus making
the CSTR heat duty to drop which temporarily decreasing tray #5 temperature causing the inverse response.
Increasing EtOH feed ow rate will not have the inverse
response because EtOH is a low-boiler thus eventually
will make tray #5 temperature to drop which coincide
with the eect of decreasing CSTR heat duty due to level
control action. Thus, the rst overall control strategy to
further dynamically evaluate the control performance is
to use CSTR heat duty to control CSTR level and to

224

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

100

ETOH feed +1%

ETOH feed -1%

92

95.4
95.2
95.0

st

st

94

95.6

95.6

T of 1 column (C)
5

T of 1 column (C)
5

96

st

T of 1 column (C)
5

95.8
98

94.8
94.6
AA feed +1%
AA feed -1%

94.4

10

15

20

25

30

95.2
95.0
94.8
Q1 +1%
Q1 -1%

94.6

94.2
0

95.4

94.4
0

10

Time (hr)

15

20

25

30

10

Time (hr)

15

20

25

30

20

25

30

Time (hr)

4.372
4.370
4.368

4.366
4.364
4.362
4.360

4.370

CSTR level (m)

4.368

CSTR level (m)

CSTR level (m)

4.370

4.366
4.364
4.362

4.365

4.360

4.360
ETOH feed +1%
ETOH feed -1%

4.358

AA feed +1%
AA feed -1%

4.358

Q1 +1%
Q1 -1%

4.355

4.356
0

10

15

20

25

30

10

Time (hr)

15

20

25

30

10

Time (hr)

15

Time (hr)

Fig. 12. Open-loop responses with step changes in either of EtOH feed, AA feed, or CSTR duty.

use EtOH feed ow rate to control tray #5 temperature

of rectier. AA feed will be used as the throughput
manipulator and other control loops are explained previously. This overall control strategy is denoted as Control
Strategy (I). This overall control strategy is similar to
what Chien et al. [1] used in their ethyl acetate system.
However, in their control strategy the tray temperature
at rectier was controlled by manipulating the acid feed
ow rate rather than the EtOH feed ow rate as is used
in our system.
Another overall control strategy from previous CS3 is to
use CSTR heat duty as throughput manipulator and letting
organic reux ratio to be xed under various disturbances.
Note again from previous Fig. 9, this CS3 results in the
least temperature deviations under perfect composition
control. In this CS3, AA and EtOH feed ow rates will
be used as the manipulated variables for the CSTR level
loop and tray #5 temperature control loop at rectier.
The loop pairing selection is to use AA feed to control
CSTR level and to use EtOH feed to control tray #5 temperature at rectier since the eect of AA feed to this tray
temperature is considerably smaller as can be seen in previous Fig. 12. This second overall control strategy using two
feed ow rates to control CSTR level and one tray temperature at rectier is denoted as Control Strategy (II). Kaymak and Luyben [14] in their reactive distillation column
control study proposed to use two feed ow rates to control two tray temperatures at the reactive distillation column. Although their control strategy is dierent than our
study, the concept of using heat duty as throughput manipulator is the same.

4.3. Open-loop sensitivity analysis to verify the suitability

of temperature control points
Before the dynamic evaluation of closed-loop performances for the two overall control strategies, the open-loop
sensitivity analysis will be performed to verify that the temperature control point determined by closed-loop sensitivity analysis in Section 4.1 is really workable or not. In
order to have the tray temperature control loop to work
well, the chosen manipulated variable needs to have
enough open-loop sensitivity to the controlled tray temperature. For the open-loop sensitivity plots of Control Strategy (I), CSTR level loop is in automatic mode manipulated
by the CSTR heat duty. Similarly for the open-loop sensitivity plots of Control Strategy (II), the CSTR level loop is
in automatic mode manipulated by AA feed ow rate. Similar open-loop sensitivities are observed for either of Control Strategies (I) or (II). Fig. 14 shows the example
result for Control Strategy (II). From this gure, it is
shown that enough open-loop sensitivity is observed for
the selected temperature control points. The same conclusion can be made for Control Strategy (I).
4.4. Closed-loop dynamic simulation results
Rigorous dynamic simulator, Aspen DynamicsTM, was
used for all the closed-loop dynamic simulations. The more
dicult to handle unmeasured feed composition changes
will be tested rst on Control Strategies (I) and (II).
Fig. 15 shows the closed-loop dynamic simulation with
AA feed concentration changes from totally AA to include

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

225

120

CSTR Level <-> Q1 in auto mode

96.5

Temperature ( oC)

110

95.5

95.0

st

T5 of 1 Column (C)

96.0

94.5

90

AA Feed +1%
AA Feed -1%

94.0

100

Base case
EtOH feed +0.01%
EtOH feed -0.01%

93.5
0

10

15

20

25

30

80

Time (hr)

1 2

9 10 11 12 13 14 15 16 17 18 19 20

Trays of 1ST Column

CSTR Level <-> Q1 in auto mode

99

105
Base case
Q2 +1%
Q2 -1%

98
97

Temperature ( oC)

96

st

T5 of 1 Column (C)

100

95
94

95

90

85

EtOH Feed +1%

EtOH Feed -1%

93

80

92
0

10

15
Time (hr)

20

25

30

Fig. 13. Open-loop responses for tray #5 temperature of rectifying

column with step changes in either AA feed or EtOH feed while CSTR
level manipulating its duty is in auto mode.

75
1

Trays of 2nd Column

Fig. 14. Open-loop sensitivity plot for the rectifying and stripping
columns while CSTR level is controlled by AA feed.

5% water at t = 2 h. The three important control loops

(tray #5 temperature at rectier, tray #5 temperature at
stripper, and CSTR level) are all controlled back to their
setpoint values. The rather slowness of the dynamic
response is mainly due to the needed large volume of the
CSTR reactor in the optimized owsheet. With a smaller
CSTR reactor, the stringent AA impurity specication in
the nal product can not be satised. Since interactions
between the CSTR level loop and rectier T5 loop are quite
severe as can be seen in previous Fig. 12, together with the
unusual inverse response characteristics in the o-diagonal
element (rectier T5 vs. Q1), the pattern of the closed-loop
dynamic response is rather unusual.
By looking at the nal product composition, although
the EtOH impurity in the nal product stream holds nicely
using this control strategy, however, there is some deviation
of the AA concentration with this impurity changing from
0.005 wt% to 0.0052 wt%. This agrees with the closed-loop
sensitivity plot in previous Fig. 7 where ideally the temperature setpoint at rectier should be adjusted a little in the
face of AA feed composition disturbance.

Fig. 16 shows the simulation results for 20% changes

in the EtOH feed composition (water contents in this feed
stream from 17.8 mol% to 21.35 mol% or from 17.8 mol%
to 14.24 mol%) at t = 2 h. The control strategy performs
well for this type of disturbance with both the AA and
EtOH impurities in the product stream holding very close
to their specications.
The closed-loop responses of these two types of feed
composition changes for Control Strategy (II) are shown
in Figs. 17 and 18. It is noticed that for Control Strategy
(II) both AA and EtOH impurities are very close to their
specications despite various feed composition changes.
This agrees with the earlier ndings via steady-state simulation (see Fig. 9) where no adjustment of the temperature
setpoints are needed for AA or EtOH feed composition
changes.
It is also noticed that the transient response also favor
Control Strategy (II) in that the ranges of variations in
the two product impurities during the transient period
are much narrower for Control Strategy (II). For example,

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

Final product EtOH Concentration

226

Final product AA Concentration

6.0e-5
5.5e-5

5.20e-5

5.0e-5
4.5e-5
4.0e-5
3.5e-5
3.0e-5
2.5e-5

50

100

150

200

250

1.015e-3
1.010e-3

1.007e-3

1.005e-3
1.000e-3
9.950e-4
9.900e-4

300

50

100

95

94

93

92

250

300

150
200
Time (hr)

250

300

200

250

300

200

250

300

102
100
98
96
94

50

100

150

200

250

300

50

100

50

100

Reboiler Duty of 2nd Column (KW)

Time (hr)
99.645

T5 of 2nd Column ( o C)

200

104

EtOH Feed Rate (mol/min)

T5 of 1st Column (o C)

96

99.640
99.635
99.630
99.625
99.620
99.615

50

100

150

200

250

72
70
68
66
64
62
60

300

Reboiler Duty of 1st Column (KW)

4.365
4.364
4.363
4.362
4.361
4.360

50

100

150

150

Time (hr)

Time (hr)

CSTR Level (m)

150

Time (hr)

Time (hr)

200

250

300

Time (hr)

154
152
150
148
146
144
142
140

50

100

150

Time (hr)
AA Concentration-5%

Fig. 15. Closed-loop performance for Control Strategy (I) with AA feed concentration changes from 100 mol%AA to 95 mol%AA at t = 2 h.

with 20% changes in the EtOH feed composition, the

range of variations in AA impurity is from 0.0040 wt% to
0.0063 wt% for Control Strategy (II). However, much larger range of variations is observed for Control Strategy
(I) with range from 0.0028 wt% to 0.0086 wt%.
The controller tuning for either control strategy follows
the same tuning rules. The CSTR level control loop is
tuned rst with IMC-PID tuning method of Chien and
Fruehauf [15] using integrating plus deadtime model. The
closed-loop time constant is set to be twice of the model

apparent deadtime. For the Control Strategy (II), since

the CSTR level is controlled by the AA feed rate which
have direct inuence on the level, the tuning is tighter than
the result of Control Strategy (I). However, by looking
at the dynamic response of AA feed rate in Fig. 17,
although the tuning is tight, the manipulated variable is
quite acceptable with small overshoot. After this important
level control loop is put into automatic mode, the temperature loop at stripper is tuned next followed by the tuning
of the temperature loop at rectier. The dynamics of the

9e-5
8e-5
7e-5
6e-5

5.03e-5
5e-5

4.90e-5
4e-5
3e-5
2e-5

50

100

150

200

250

Final product EtOH Concentration

Final product AA Concentration

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

1.015e-3
1.010e-3
1.005e-3

1.002e-3

1.000e-3

9.974e-4

9.950e-4
9.900e-4
9.850e-4

300

50

100

99

EtOH Feed Rate (mol/min)

T5 of 1st Column ( oC)

200

250

300

200

250

300

200

250

300

200

250

300

110

98
97
96
95
94
93
92
91

108
106
104
102
100
98
96
94

50

100

150

200

250

300

50

100

Reboiler Duty of 2nd Column (KW)

99.65
99.64
99.63
99.62
99.61
99.60
99.59

50

100

150

200

250

74
73
72
71
70
69
68
67

300

50

100

Reboiler Duty of 1st Column (KW)

4.370
4.368
4.366
4.364
4.362
4.360

50

100

150

150

Time (hr)

Time (hr)

150

Time (hr)

Time (hr)

T5 of 2nd Column ( oC)

150

Time (hr)

Time (hr)

CSTR Level (m)

227

200

250

300

Time (hr)

158
156
154
152
150
148
146

50

100

150

Time (hr)
EtOH Feed H2O Concentration +20%
EtOH Feed H2O Concentration -20%

Fig. 16. Closed-loop performance for Control Strategy (I) with EtOH feed concentration 20% changes at t = 2 h.

temperature loop at stripper is much faster than the one at

rectier, thus it is tuned rst and put into automatic mode
before the tuning of the nal temperature loop at rectier.
Integrating plus deadtime model was also used with the
closed-loop time constant to be twice of the model apparent deadtime. The tuning of this nal temperature control
loop is crucial for the overall control performance. The
suitable combination of the controller gain and reset time

is important to achieve faster closed-loop response without

much of the oscillation.
Since Control Strategy (II) works better to handle the
unmeasured feed composition disturbances, this control
strategy is further tested with simulation study for the case
with throughput changes. For this control strategy,
increasing or decreasing the production rate of the nal
product needs to go through the changes of the CSTR heat

Final product EtOH Concentration

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

Final product AA Concentration

228
5.04e-5
5.02e-5

5.02e-5

5.00e-5
4.98e-5
4.96e-5
4.94e-5
4.92e-5
4.90e-5
4.88e-5
4.86e-5
4.84e-5

50

100

150

200

250

1.004e-3
1.003e-3

1.003e-3

1.002e-3
1.001e-3
1.000e-3
9.990e-4
9.980e-4

300

50

100

150

200

250

300

200

250

300

200

250

300

200

250

300

Time (hr)

Time (hr)

EtOH Feed Rate (mol/min)

T5 of 1st Column ( oC)

95.10
95.05
95.00
94.95
94.90

102.0
101.8
101.6
101.4
101.2
101.0
100.8
100.6
100.4

94.85

50

100

150

200

250

300

50

100

Reboiler Duty of 2nd Column (KW)

T5 of 2nd Column (oC)

99.65
99.64
99.63
99.62
99.61
99.60
99.59

50

100

150

200

250

70.5
70.0
69.5
69.0
68.5
68.0
67.5
67.0

300

50

100

69.5

AA feed Rate (mol/min)

4.36435
4.36430

CSTR Level (m)

150

Time (hr)

Time (hr)

4.36425
4.36420
4.36415
4.36410
4.36405
0

150

Time (hr)

Time (hr)

69.0
68.5
68.0
67.5
67.0
66.5
66.0
65.5

50

100

150

200

250

300

Time (hr)

50

100

150

Time (hr)
AA Concentration-5%

Fig. 17. Closed-loop performance for Control Strategy (II) with AA feed concentration changes from 100 mol%AA to 95 mol%AA at t = 2 h.

duty. Fig. 19 shows the simulation results with 10%

changes of the CSTR heat duty. With these changes of
the CSTR heat duty, the two feed ow rates are changed
accordingly with only small deviations in the AA product
impurity. Because this disturbance is considered as a
known load change, thus similar to the paper by Huang
et al. [13], a planning of the temperature setpoint adjustment versus each CSTR heat duty can easily be made from
process simulation to circumvent the impurity deviation

problem. For example, with 10% changes in the throughput manipulator (CSTR heat duty), the setpoint of tray #5
temperature at rectier should decrease around 0.3 C
according to Fig. 9 so that AA impurity in the nal product
can return back to 0.005 wt%.
With above 10% changes of the CSTR heat duty, the
product rate is changed from 63.8824 mol/min to
69.66846 mol/min (a +9.1% increase in production) or
from 63.8824 mol/min to 58.0364 mol/min (a 9.1%

6.5e-5

Final product EtOH Concentration

Final product AA Concentration

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

6.0e-5
5.5e-5
4.98e-5

5.0e-5

4.95e-5

4.5e-5
4.0e-5
3.5e-5
0

50

100

150

200

250

1.015e-3
1.010e-3
1.00e-3

1.005e-3
1.000e-3
9.950e-4

9.96e-4

9.900e-4
9.850e-4
0

300

50

100

96.5

EtOH Feed Rate (mol/min)

T5 of 1st Column ( oC)

200

250

300

200

250

300

200

250

300

200

250

300

112

97.0

96.0
95.5
95.0
94.5
94.0

110
108
106
104
102
100
98
96
94
92

93.5
0

50

100

150

200

250

300

50

100

150

Time (hr)
Reboiler Duty of 2 Column (KW)

Time (hr)
99.640
99.635
99.630
99.625

73
72
71

nd

o
nd
T5 of 2 Column ( C)

150

Time (hr)

Time (hr)

99.620
99.615
99.610
99.605
99.600
0

50

100

150

200

250

70
69
68
67
0

300

50

100

150

Time (hr)

Time (hr)
69

AA Feed Rate (mol/min)

4.3645
4.3644

CSTR Level (m)

229

4.3643
4.3642
4.3641

68
67
66
65
64
63

4.3640
0

50

100

150

200

250

300

Time (hr)

50

100

150

Time (hr)
EtOH Feed H2O Concentration +20%
EtOH Feed H2O Concentration -20%

Fig. 18. Closed-loop performance for Control Strategy (II) with EtOH feed concentration 20% changes at t = 2 h.

decrease in production). This demonstrates that CSTR

heat duty can be used as a throughput manipulator to
smoothly increase or decrease the production rate of the
nal product. The nal proposed overall control strategy
for this process can be seen in Fig. 20.
5. Conclusions
In this paper, design and control of a complete ethyl
acrylate process with coupled reactor/columns congura-

tion has been investigated. Unlike other paper in the literature with only one reactive distillation column to
produce products like methyl acetate or butyl acetate, this
overall process owsheet is more complex including a
CSTR, a rectifying column, a decanter, and another stripping column. The design procedure is based on the maximization of total annual prot (TAP) for the overall
system. The nal EA product is having very high purity
of 99.78 mol% (or 99.89 wt%) with two impurities of
AA and EtOH meeting stringent specications of

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

7.0e-5

Final product EtOH Concentration

Final product AA Concentration

230

6.5e-5
6.0e-5
5.5e-5

5.24e-5

5.0e-5
4.77e-5

4.5e-5
4.0e-5
3.5e-5
0

50

100

150

200

250

1.08e-3
1.06e-3
1.04e-3
1.01e-3

1.02e-3
1.00e-3
9.80e-4

9.89e-4

9.60e-4
9.40e-4
9.20e-4
0

300

50

100

97
96
95
94
93

250

300

200

250

300

200

250

300

200

250

300

115
110
105
100
95
90
85

50

100

150

200

250

300

50

100

Reboiler Duty of 2nd Column (KW)

99.75
99.70
99.65
99.60
99.55
99.50
0

50

100

150

150

Time (hr)

Time (hr)

T5 of 2nd Column ( oC)

200

120

EtOH Feed Rate (mol/min)

T5 of 1st Column ( oC)

98

200

250

80
78
76
74
72
70
68
66
64
62
60
0

300

50

100

150

Time (hr)

Time (hr)
76

AA Feed Rate (mol/min)

4.368

CSTR Level (m)

150

Time (hr)

Time (hr)

4.366

4.364

4.362

74
72
70
68
66
64
62
60
58
56

50

100

150

200

250

300

Time (hr)

50

100

150

Time (hr)
Throughput +10%
Throughput-10%

Fig. 19. Closed-loop performance for Control Strategy (II) with 10% changes in CSTR duty at t = 2 h.

0.005 wt% AA and 0.1 wt% EtOH in the nal product

stream.
For the overall control strategy of this process, a
straightforward procedure has been followed which uses
closed-loop sensitivity analysis to screen out the worse
candidates of overall control strategy and to pick the
suitable temperature control points. The nal recommended overall control strategy is found to be dierent
than another similar coupled reactor/columns process pub-

lished earlier [1]. Both AA and EtOH feed ow rates are

used as manipulated variables in the overall control strategy to control CSTR level and one tray temperature at rectier with CSTR heat duty left as the throughput
manipulator for the overall process. The nal EA product
with stringent specications of EtOH and AA impurities
can be achieved with this proposed overall control strategy
despite various feed composition disturbances and
throughput changes.

I-L. Chien et al. / Journal of Process Control 18 (2008) 215231

231

Water
FC
TC

TC
X

Decanter
LC

PC

LC

Organic
Reflux

Aqueous

PC

1
FC

TC

X
FC

AA
TC

FC

19

EtOH

LC

Steam
7

20
LC

CSTR

Steam (throughput manipulator)

Reboiler
EA

Fig. 20. Proposed overall control strategy of this process.

Acknowledgements
The rst author, I-Lung Chien, take this opportunity to
thank Prof. Dale Seborg for his guidance and valuable advices over the years. This paper was prepared while I-Lung
was back visiting Department of Chemical Engineering
UCSB after completing his Ph.D. degree as Dales student
more than 20 years ago. There is an old Chinese saying
which conveys I-Lungs appreciation to Prof. Seborg:
Once becoming a teacher of yours, always respect him
as your father. At this occasion of Dales 65th birthday,
I-Lung wish him happiness always and longevity.
References
[1] I-L. Chien, Y.P. Teng, H.P. Huang, Y.T. Tang, Design and control of
an ethyl acetate process: coupled reactor/column conguration, J.
Proc. Cont. 15 (2005) 435449.
[2] M. Witczak, W. Grzesik, J. Skrzypek, Kinetyka estrykacji kwasu
akrylowego nizszymi alkoholami alifatycznymi, Inz. Chem. Procesowa 25 (2004) 331340.
[3] C.K. Yi, W.L. Luyben, Design and control of coupled reactor/
column systems Part 1. A binary coupled reactor/rectier system,
Comput. Chem. Eng. 21 (1997) 2546.
[4] C.K. Yi, W.L. Luyben, Design and control of coupled reactor/
column systems Part 2. More complex coupled reactor/column
system, Comput. Chem. Eng. 21 (1997) 4767.

[5] C.K. Yi, W.L. Luyben, Design and control of coupled reactor/
column systems Part 3. A reactor/stripper with two columns and
recycle, Comput. Chem. Eng. 21 (1997) 6986.
[6] S.F. Chiang, C.L. Kuo, C.C. Yu, D.S.H. Wong, Design alternatives
for the amyl acetate process: coupled reactor/column and reactive
distillation, Ind. Eng. Chem. Res 41 (2002) 32333246.
[7] U. Weidlich, J. Gmehling, A modied UNIFAC model. 1. Prediction
of VLE, HE and c1, Ind. Eng. Chem. Res. 26 (1987) 13721381.
[8] J. Gmehling, J. Li, M. Schiller, A modied UNIFAC model. 2.
Present parameter matrix and results for dierent thermodynamic
properties, Ind. Eng. Chem. Res. 32 (1993) 178193.
[9] J.G. Hayden, J.P. OConnell, A generalized method for predicting
second virial coecients, Ind. Eng. Chem. Proc. Des. Dev. 14 (1975)
209216.
[10] J.M. Douglas, Conceptual Process Design, McGraw-Hill, New York,
1988.
[11] R. Turton, R.C. Bailie, W.B. Whiting, J.A. Shaeiwitz, Analysis,
Synthesis, and Design of Chemical Processes, Prentice-Hall, Englewood Clis, NJ, 1998.
[12] H.Y. Lee, H.P. Huang, I-L. Chien, Control of reactive distillation
process for production of ethyl acetate, J. Proc. Cont. 17 (2007) 363
377.
[13] S.G. Huang, C.L. Kuo, S.B. Huang, Y.W. Chen, C.C. Yu, Temperature control of heterogeneous reactive distillation, AIChE J. 50
(2004) 22032216.
[14] D.B. Kaymak, W.L. Luyben, Comparison of two types of twotemperature control structures for reactive distillation columns, Ind.
Eng. Chem. Res. 44 (2005) 45254640.
[15] I-L. Chien, P.S. Fruehauf, Consider IMC tuning to improve
controller performance, Chem. Eng. Prog. 86 (1990) 3341.