-
Available online at
www.sciencedirect.comwww.elsevier.com/locate/jprocont
Journal of Process Control 18 (2008) 215231Overall control
strategy of a coupled reactor/columnsprocess 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 2007Abstract
Ethyl acrylate (EA) is widely used in industry as a precursor
for varnishes, adhesive, and finishes of papers and textiles. This
impor-tant ester can be produced directly from ethanol (EtOH) and
acrylic acid (AA) via esterification reaction with the presence of
sulphuricacid as homogeneous catalyst. The proposed design
flowsheet of this process includes a CSTR reactor coupled with a
rectifier and anoverhead decanter. In order to further purify the
final EA product, another stripper is needed with its top vapor
recycled back to decan-ter. The simplest and industrial easily
applicable overall control strategy will be investigated with only
one tray temperature control loopin each of the two columns. The
final proposed overall control strategy of this process is found to
be different than another similar cou-pled reactor/columns process
published earlier [I-L. Chien, Y.P. Teng, H.P. Huang, Y.T. Tang,
Design and control of an ethyl acetateprocess: coupled
reactor/column configuration, J. Proc. Cont. 15 (2005) 435449].
Both EtOH and AA feed flow rates are used as manip-ulated variables
in the overall control strategy with CSTR heat duty left as
throughput manipulator for the overall process. The final EAproduct
with stringent specifications of 0.1 wt% EtOH and 0.005 wt% AA
impurities can be achieved with this proposed overall
controlstrategy despite feed flow rate and feed composition
disturbances. 2007 Elsevier Ltd. All rights reserved.
Keywords: Ethyl acrylate; Esterification reaction; Reactive
distillation; Coupled reactor/columns; Optimum design; Overall
control strategy1. Introduction
Ethyl acrylate (EA) is widely used in industry as a pre-cursor
for varnishes, adhesive, and finishes of papers andtextiles. This
important ester can be produced directly fromethanol (EtOH) and
acrylic acid (AA) via esterificationreaction with the presence of
sulphuric acid as homoge-neous catalyst. Only in a very recent
paper [2], the kineticsof this esterification reaction has been
given. There is nopaper in the literature on the subject of the
production ofethyl acrylate, thus the results of this paper should
be use-ful to other researchers.
In this study, overall control strategy of this processwith
coupled reactor/columns configuration will be stud-0959-1524/$ -
see front matter 2007 Elsevier Ltd. All rights
reserved.doi:10.1016/j.jprocont.2007.02.006
* Corresponding author. Tel.: +886 2 27376652; fax: +886 2
27376644.E-mail address: [email protected] (I-Lung Chien).ied.
The principal behind the coupled reactor/columns con-figuration,
similar to reactive distillation, is that thecontinuous removal of
products from the esterificationreaction mixture by distillation
reduces the backward reac-tion rate. The advantage of the coupled
reactor/columnsconfiguration over reactive distillation according
to Yiand Luyben [3] include: the existing reactor/columns inthe
plant can be retrofitted for this usage; easy maintenanceof the
overall system; larger reactor holdup and differentreaction
temperature can easily be designed; etc.
In a three-paper series by Yi and Luyben [35], theystudied the
design and control of various coupled reac-tor/column systems. The
studied systems include: a binaryreactor/rectifier, a binary
reactor/stripper, a multicompo-nent reactor/rectifier, a
multicomponent reactor/rectifier/stripper, and a more complex
process that consists of acoupled reactor/stripper, two
distillation columns andone recycle stream. Their studied systems
are very simple,
mailto:[email protected]
-
216 I-L. Chien et al. / Journal of Process Control 18 (2008)
215231ideal chemical systems and also no liquidliquid equilib-rium
is considered. Chiang et al. [6] studied a coupled reac-tor/column
system for the production of amyl acetate.Their process is much
simpler than the studied ethyl acry-late process because amyl
acetate has the highest boilingpoint in the system. Their system
with the configurationof reactor with rectifier on top and stripper
on the bottomproduces amyl acetate from the bottom of the stripper
andalso produces water through aqueous phase of a decanter.Chien et
al. [1] proposed a coupled reactor/column config-uration for the
production of ethyl acetate. Their processflowsheet configuration
is very similar to the ones in thispaper. However, the overall
control strategy is differentthan the one will be developed for
this system.
The organization of this paper is as follows. The ther-modynamic
properties of this four-component system andthe kinetics of this
esterification reaction will be given inSection 2. The design
flowsheet of a complete coupled reac-tor/column system will be
proposed in Section 3. A finalEA product purity of over 99.5 wt%
will be obtained inthe proposed design with stringent
specifications of0.1 wt% EtOH and 0.005 wt% AA impurities. In
Section4, the overall control strategy of this process will be
inves-tigated. Only one tray temperature control loop in each ofthe
two columns (rectifier and stripper) will be used. Feedflow rate
and feed composition disturbances will be usedto test the overall
control strategy. Some concludingremarks will be drawn in the final
section.2. Thermodynamic and kinetic model used in the
simulation
There are total of four azeotropes in this system includ-ing two
homogeneous azeotropes of EtOH + H2O andEtOH + EA and two
heterogeneous azeotropes ofEA + H2O and EtOH + EA + H2O. In order
to accuratelyrepresent the overall system, liquid activity
coefficientmodel was used for the vaporliquidliquid equilibrium.A
suitable NRTL (nonrandom two-liquid) model parame-ter set has been
established with excellent prediction of thecompositions and
temperatures for the four azeotropes inthis system. In this NRTL
parameter set, the Aspen Plus
built-in binary-pair parameters of AAEA and AAH2Owere used. For
the EtOHH2O, EtOHEA, EAH2O pairs,binary parameters were obtained to
fit well the LLEboundary of these three components. For the one
pair ofAAEtOH that does not have the Aspen Plus built-inNRTL binary
parameters, the Dortmund modified UNI-FAC group contribution
estimation method [7,8] was usedto obtain the remaining
thermodynamic model parameters.Vapor association of Acrylic acid
due to dimerization hasalso been included by using the second
virial coefficientof the HaydenOConnell [9] model in the vapor
phase.The Aspen Plus built-in association parameters wereemployed
to compute fugacity coefficient.
The kinetic model of this esterification reaction is fromthe
paper by Witczak et al. [2] with diluted sulphuric acidas
homogeneous catalyst. The reaction can be seen asbelow:
C2H5OHEtOH
CH2CHCOOHAA
$ CH2CHCOOC2H5EA
H2O 1
The kinetic equation is
r k1C cat C2AAC2EtOH C2EAC
2H2O
K2
!mol=dm3 min
where
k1 3:26 106 exp15900RT
dm12=mol4 min
K 2:71 104 exp 6490RT
and R is gas constant (1.987 cal/mol/K) and Ccat is as-sumed to
have value of 0.15 mol/dm3. In the kinetic equa-tion, all
concentrations are with unit of (mol/dm3) andtemperature in K.
3. Design flowsheet of the complete process
The RCM for the EtOHEAH2O three-componentsystem and AAEAH2O
three-component system can beseen in Figs. 1 and 2. From these two
figures, the highestboiling point temperature of the whole system
includingthe pure components and azeotropes is the acrylic acid(AA)
at 141.19 C and the lowest temperature of the wholesystem is the
EtOH + EA + H2O three-component azeo-trope at 75.29 C. The two
products (ethyl acrylate andwater) of this esterification reaction
are neither the lightestnor the heaviest component in the system,
thus the com-plete designed process will need to be more complex
incomparison with the other reactive distillation papers inthe
literature.
The proposed design of this process including a CSTRreactor
coupled with a rectifier (without heat source). Theheat input in
the CSTR totally vaporizes the reactor outletstream to vapor phase
and continuously enter the rectifierfrom the bottoms to promote the
forward reaction further.The bottom liquid stream from the
rectifier containingmostly heavy boiler AA is recycled back to the
CSTR.The composition of the top vapor stream from the rectifieris
close to the lightest boiler of ternary azeotrope of EtO-H + EA +
H2O. This stream after sub-cooling to 40 Ccan be naturally
separated inside a decanter to formorganic and aqueous phases.
Extra water is added in thedecanter to maintain suitable
composition inside of theliquidliquid boundary. The water purity of
the aqueousphase is quite high, thus it can go to a waste water
treat-ment plant for discharge. The organic phase compositionby
natural liquidliquid separation has the benefit of cross-ing the
distillation boundary into a desirable region toobtain pure ethyl
acylate product (see Fig. 1). This organic
-
EA WATER
AA
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
(141.19 C)
(99.40 C)(81.04 C)
(100.02 C)
Fig. 2. RCM of AAEAH2O three-component system.
EA WATER
EtOH
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9 (78.20 C)
(77.87 C)
(99.40 C)
(78.31 C)
(100.02 C)(81.04 C)
(75.29 C )
Fig. 1. RCM of EtOHEAH2O three-component system.
I-L. Chien et al. / Journal of Process Control 18 (2008) 215231
217phase stream is partly refluxed and is partly designed tofeed
into another stripper with reboiler for further purifica-tion into
the final EA product. The top vapor of this strip-per with
composition near the top vapor of the rectifier is
-
Organic Reflux
AA
Steam
Aqueous
EtOH
Decanter
Steam
EA
Reboiler
Water
CSTR
Fig. 3. Conceptual design of the overall process flowsheet.
218 I-L. Chien et al. / Journal of Process Control 18 (2008)
215231also condensed and then fed into the decanter. The
bottomstream of the stripper is the final EA product with
stringentspecifications of 0.1 wt% EtOH and 0.005 wt% AA
impuri-ties in this product stream. This conceptual design of
theoverall process flowsheet can be seen in Fig. 3.
The design flowsheet is selected based on the maximiza-tion of
Total Annual Profit (TAP) for the overall system.This TAP includes:
the product value minus the costs oftwo feed streams, minus
annualized capital costs, minustotal utility costs, and minus the
waste water treatmentColumn 1=24 stagesColumn 2=7 stagesCSTR
volume=126 cum
Water flow rate (mol/min)0 20 40 60 80 100 120
Prof
it ($
/yea
r)
5.25e+5
5.30e+5
5.35e+5
5.40e+5
5.45e+5
5.50e+5
5.55e+5
5.60e+5
Water flow rate=50 mole/minCSTR volume=126 cumColumn 2=7
stages
Tray of Column 122 23 24 25 26
Prof
it ($
/yea
r)
5.562e+5
5.563e+5
5.564e+5
5.565e+5
5.566e+5
5.567e+5
5.568e+5
Fig. 4. Optimization result with EtOcost. The reason to add
product value and the costs oftwo feed streams in the TAP
calculation is because thetwo feeds are not necessarily fixed at
equal molar ratio.The annualized capital costs follow directly from
the calcu-lation procedure in Douglas [10] with the annual
capitalcharge factor of 1/3 was used. The annualized equipmentcost
includes CSTR, column shells, column trays, reboilers,and
condensers. The utility cost including the steam andcooling water
costs are calculated the same way as inChiang et al. [6]. The waste
water treatment cost is calcu-Water flow rate=50 mole/minColumn
1=24 stagesColumn 2=7 stages
CSTR Volume (cum)125 126 127 128 129 130 131 132 133
Prof
it ($
/yea
r)
5.46e+5
5.48e+5
5.50e+5
5.52e+5
5.54e+5
5.56e+5
5.58e+5
Water flow rate=50 mole/minCSTR volume=126 cumColumn 1=24
stages
Tray of Column 25 9
Prof
it ($
/yea
r)
5.561e+5
5.562e+5
5.563e+5
5.564e+5
5.565e+5
5.566e+5
5.567e+5
5.568e+5
6 7 8
H feed flow rate of 96 mol/min.
-
Table 1Optimum process design for this system
CSTR holdup (m3) 64
Rectifier total stages(where 20th stage is the CSTR)
20
Stripper total stages(including the reboiler)
7
Fresh EtOH feed flow rate(g mol/min)
102
EtOH feed composition 82.2 mol% EtOH17.8 mol% H2O
Fresh AA feed flow rate(g mol/min)
66
AA feed composition 100 mol% AAVapor flow rate from CSTR
to rectifier (g mol/min)258
Bottom liquid flow rate from rectifier toCSTR (g mol/min)
90
CSTR heat duty (KW) 151.8Water injection rate into the
decanter (g mol/min)100.0
Organic reflux flow rate (g mol/min) 84.05Organic outlet flow
rate into
stripper (g mol/min)145.55
Stripper reboiler duty (KW) 70.10Aqueous outlet flow rate (g
mol/min) 204.1EA product flow rate (g mol/min) 63.9
EA product composition 99.78 mol% (99.89 wt%) EA0.0069 mol%
(0.005 wt%) AA0.213 mol% (0.10 wt%) EtOH1 104 mol% H2O
I-L. Chien et al. / Journal of Process Control 18 (2008) 215231
219lated with the estimation of $0.053/m3 (given in Table 3.4of
Turton et al. [11] textbook).
The design and operating variables that need to bedetermined
include AA/EtOH feed ratio, CSTR holdup,total stages of the
rectifier, total stages of the stripper,and the water addition rate
into the decanter. An iterativesequential optimization procedure is
proposed to find theoptimal flowsheet of the overall system. In the
design ofthe process flowsheet, pure AA feed composition isassumed
while the EtOH feed stream is practically assumedto contain 82.2
mol% EtOH and 17.8 mol% H2O.
The acrylic acid flow rate is fixed at 66 mol/min to setthe
throughput of the overall process. The organic refluxflow rate from
the decanter is manipulated to hold theAA impurity in the final EA
product to be at the specifica-tion of 0.005 wt%. The reboiler duty
of the stripper ismanipulated to hold the EtOH impurity in the
final EAproduct to be at the specification of 0.1 wt%. The
iterativesequential optimization procedure is to find the
optimiza-tion result at a particular EtOH feed flow rate. In
eachoptimization search, four design and operating variablescan be
changed including CSTR holdup volume, rectifyingcolumn total stage,
stripping column total stage, and wateraddition rate. Iterative
procedure is needed to find the opti-mization result. Fig. 4 shows
an example of the optimiza-tion result with EtOH feed flow rate at
96 mol/min. Inthis case, the water addition rate to achieve the
highestTAP is at 50 mol/min, the CSTR holdup volume is126 m3,
rectifying column total stage is 24, and strippingcolumn total
stage is 7. Notice that in this case, CSTRholdup volume below 126
m3 cannot satisfy the AA impu-rity specification in final EA
product stream.
Summarizing the optimization results at various EtOHfeed flow
rates, Fig. 5 shows that the optimal EtOH feedflow rate is at 102
mol/min. At this flow rate, the TAP ismaximized at $5.95 105. The
optimum pure AA and pureEtOH feed ratio is calculated to be 1:1.27,
not at exactlyequal molar ratio. With this optimum feed ratio,
otheroptimum design and operating variables are: water addi-EtOH
flow rate (mol/min)94 96 98 100 102 104 106 108 110
Prof
it ($
/yea
r)
5.5e+5
5.6e+5
5.7e+5
5.8e+5
5.9e+5
6.0e+5
Fig. 5. Summary of optimization results at various EtOH feed
flow rates.tion rate at 100 mol/min, CSTR holdup volume at 64
m3,rectifying column total stage at 20, and stripping columntotal
stage at 7. The final overall process flowsheet can beseen in Table
1. Notice from the final flowsheet, the speci-fications of the two
impurities are met and the final EAproduct purity is at 99.78 mol%
(or 99.89 wt%).
4. Overall control strategy of this process
Some straightforward inventory and other loops aredetermined
first. These include the following: the organicphase level is
controlled by the organic outlet flow tostripper; the aqueous phase
level is controlled by the aque-ous outlet flow; the bottom level
of the stripper is con-trolled by the final product flow; the top
pressures ofthe rectifier and stripper are controlled at 1.1 atm by
thetop vapor flow; and the temperature at decanter are con-trolled
at 40 C by the condenser duty. The organic refluxratio is fixed by
a ratio scheme where the value of theratio can be set by a tray
temperature control loop if nec-essary. The extra water flow rate
into the decanter is ratioto a throughput manipulator yet to be
determined. Sum-marizing of the basic loops in the process can be
seen inFig. 6.
For the candidate overall control strategies studied inthis
paper, some other prerequisite assumptions are out-lined below.
Firstly, all control loops are in conventionalPID form for easier
industrial applications. Secondly,
-
Organic Reflux
AA
Steam
Aqueous
EtOH
Decanter
Steam
EA
Reboiler
Water
CSTR
PC PCLC LC
LC
FC
FC
FC
FC
1 1
196
207
TCTC
X
Ratio to athroughputmanipulator
Fig. 6. Basic regulatory control strategy of the overall
process.
220 I-L. Chien et al. / Journal of Process Control 18 (2008)
215231only one tray temperature control loop is investigated ineach
of the rectifier and stripper column to avoid strongloop
interactions. Thirdly, the manipulated variable ofthe tray
temperature control loop for the stripper arestraightforwardly
determined as the stripper reboilerduty.
The manipulated variables of the remaining two mostimportant
control loops are yet to be determined. Thesetwo control loops are
the CSTR level control loop andthe rectifier tray temperature
control loop. The CSTR levelcontrol loop unlike the other level
control loops needs toinclude integral mode to make sure the
reaction volumeis maintained. The candidate manipulated variables
forthese two important loops are: AA feed flow, EtOH feedflow, CSTR
heat duty, and organic reflux ratio. Thus, theplanning of the
overall control strategy is to select twoout of the above four
manipulated variables to controlCSTR level and one tray temperature
at rectifier. For theremaining two manipulated variables, one can
be used asthe throughput manipulator to set the production rate
ofthe overall process, and the other is fixed throughout vari-ous
disturbance changes.
4.1. Determine of tray temperature control point and the
control structure
Closed-loop sensitivity analysis similar to the one usedin Lee
et al. [12] will be used here to determine the traytemperature
control point at rectifier and also at stripper.The purpose for the
closed-loop sensitivity analysis is dif-ferent in this paper. In
Lee et al. [12], the control structurewas already set and the
closed-loop analysis is solely usedto determine the temperature
control point. However, inthis paper, the closed-loop sensitivity
analysis is not onlyused to determine the tray temperature control
point atrectifier and at stripper, but it is also used to screen
outthe possible worse control structures in the overall
controlstrategy.
Because there are four free manipulated variableswhich could be
used for the CSTR level control loop andrectifier tray temperature
control loop, five possible controlstructures considered are listed
below:
CS1: EtOH feed flow and CSTR heat duty are used asmanipulated
variables for the CSTR level and recti-fier tray temperature loops,
AA feed flow is used asthroughput manipulator, and organic reflux
ratio isfixed throughout various disturbance changes.
CS2: AA feed flow and CSTR heat duty are used asmanipulated
variables for the CSTR level and recti-fier tray temperature loops,
EtOH feed flow is usedas throughput manipulator, and organic reflux
ratiois fixed throughout various disturbance changes.
CS3: EtOH feed flow and AA feed flow are used as manip-ulated
variables for the CSTR level and rectifier traytemperature loops,
CSTR heat duty is used asthroughput manipulator, and organic reflux
ratio isfixed throughout various disturbance changes.
CS4: EtOH feed flow and organic reflux ratio are used
asmanipulated variables for the CSTR level and recti-fier tray
temperature loops, AA feed flow is used asthroughput manipulator,
and CSTR heat duty isfixed throughout various disturbance changes
(onlyratio to measurable AA feed flow).
CS5: AA feed flow and organic reflux ratio are used
asmanipulated variables for the CSTR level and recti-fier tray
temperature loops, EtOH feed flow is usedas throughput manipulator,
and CSTR heat duty isfixed throughout various disturbance changes
(onlyratio to measurable EtOH feed flow).
-
Trays of 2nd Column1 2 3 4 5 6 7
Del
ta te
mpe
ratu
re (
C)
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4AA-5%EtOH(H2O)+20%EtOH(H2O)-20%throughput+10%throughput-10%
Organic reflux ratio fixed & AA fixed
Trays of 1ST Column1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
19 20
Del
ta te
mpe
ratu
re (
C)
-1.5
-1.0
-0.5
0.0
0.5
1.0
Fig. 7. Closed-loop sensitivity plot with organic reflux ratio
and AA feedfixed.
I-L. Chien et al. / Journal of Process Control 18 (2008) 215231
221There is another control structure which uses CSTRheat duty and
organic reflux ratio as manipulated variablesfor the CSTR level and
rectifier tray temperature loops.However, in this case, both EtOH
feed flow and AA feedflow will be fixed, thus will be infeasible to
cope with feedcomposition changes.
The screening of the above five possible control struc-tures can
easily be made by the closed-loop sensitivityanalysis described
below using process simulation tool.In the following, two ideal
composition control loops areassumed to be present which holds the
specifications of0.1 wt% EtOH and 0.005 wt% AA impurities in the
prod-uct stream by varying two chosen manipulated variables.One of
the manipulated variables has been pre-determinedto be the stripper
reboiler duty which controls the EtOHimpurity at 0.1 wt% in the
product stream and the othermanipulated variable to hold AA
impurity at 0.005 wt%in the product stream can be varied due to
different con-trol structure. Three disturbance changes are made
inthe closed-loop simulations, they are 10% changes inthe
throughput manipulator; 20% changes in the EtOHfeed water
concentration; and the changes of water con-centration in the AA
feed from 0 mol% to 5 mol%. Forexample, the closed-loop simulation
for control structureCS1 can be achieved with: fixing AA impurity
in the prod-uct stream at 0.005 wt% by varying EtOH feed flow
rate;fixing CSTR level at original setpoint by varying CSTRheat
duty; and fixing EtOH impurity in the product streamat 0.1 wt% by
varying stripper reboiler duty. Thus, forcontrol structure CS1 in
the simulation runs, organicreflux ratio and AA feed rate are fixed
with AA feed rateonly moves when throughput manipulator changes
arerequired.
For each disturbance case, the temperature profiles ofrectifier
and stripper under perfect composition controlcan be obtained for
the above mentioned three disturbancechanges. By comparing the
temperature profiles for thedisturbance changes versus the base
case, the deviationsof the temperature profiles of rectifier and
stripper canbe plotted. Because this is an ideal disturbance
rejectioncondition with both product specifications hold at
theiroriginal values, one could choose a tray temperature con-trol
point for rectifier and another one for stripper andusing two tray
temperature control loops to replace theideal dual-composition
control loops. The chosen temper-ature control point for rectifier
or stripper should be theone with the least deviation of the tray
temperature.Fig. 7 shows such plot for control structure CS1
underabove mentioned three kinds of disturbance changes.The focus
of observing the least deviations of the temper-atures should be
made for the feed composition changes.The reason is because the
disturbance change for thethroughput manipulator is considered as
known distur-bance, thus, some calculating scheme to adjust the
setpointvalue of the temperature control loop as proposed inHuang
et al. [13] can easily be applied to hold the
productspecifications.From Fig. 7, the suitable choice of the tray
temperaturecontrol point for rectifier is at tray #5 and for
stripper isalso at tray #5. Notice that for rectifier, the tray #5
gavevery small deviations under 20% changes in the EtOHfeed water
concentration, however, the temperature devia-tion under 5% change
in the AA feed concentration gaveconsiderably larger temperature
deviation. From theseclosed-loop simulations, one could predicts
that by usingcontrol structure CS1 with two tray temperature
controlpoints at tray #5 of rectifier and tray #5 at stripper,
thecontrol performance should perform well under 20%changes in the
EtOH feed water concentration. However,for 5% changes in the AA
feed concentration there willbe some deviations in final product
impurity especiallyfor AA impurity. Of course the chosen tray
temperaturecontrol points should be evaluated so that enough
open-loop sensitivity is there between the manipulated variable(in
this case EtOH feed rate) and rectifier tray #5 temper-ature and
also between stripper reboiler duty and stripper
-
Organic reflux ratio fixed & Q1 fixed
Trays of 1ST Column1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
19 20
Del
ta te
mpe
ratu
re (
C)
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
0.15
0.20
AA-5%EtOH(H2O)+20%EtOH(H2O)-20%throughput+10%throughput-10%
222 I-L. Chien et al. / Journal of Process Control 18 (2008)
215231tray #5 temperature. This open-loop sensitivity
evaluationwill be shown later in this section.
Similar closed-loop sensitivity simulations can be runfor other
control structures CS2, CS3, CS4, and CS5. Figs.811 show the
temperature deviations for control struc-tures CS2CS5,
respectively. By observing these plots,the tray temperatures for
various control structures withthe least temperature deviation for
rectifier 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 devia-tion
are all at tray #5. The control at stripper should beeasy and
should give acceptable control performancebecause the temperature
deviations at tray #5 under vari-ous disturbances are all not very
large. On the other hand,the control at rectifier is more difficult
and should drawmore attention. The control structure CS3 gave the
leasttemperature deviations at tray #5 of rectifier under vari-ous
disturbance changes while CS4 and CS5 gave quitelarge temperature
deviations at the least temperature devi-ation of tray #4 under 5%
change in the AA feedconcentration.Organic reflux ratio fixed &
EtOH fixed
Trays of 1ST Column1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
19 20
Del
ta te
mpe
ratu
re (
C)
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Trays of 2nd Column
1 2 3 4 5 6 7
Del
ta te
mpe
ratu
re (
C)
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4AA-5%
EtOH(H2O)+20%EtOH(H2O)-20%throughput+10%throughput-10%
Fig. 8. Closed-loop sensitivity plot with organic reflux ratio
and EtOHfeed fixed.
Trays of 2nd Column1 2 3 4 5 6 7
Del
ta te
mpe
ratu
re (
C)
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
Fig. 9. Closed-loop sensitivity plot with organic reflux ratio
and CSTRduty fixed.From the above simply closed-loop sensitivity
analysis,we can abandon further dynamic investigation of
controlstructures CS4 and CS5 and focus on CS1 to CS3 first.Only if
dynamic behaviors of CS1 to CS3 are not accept-able, we will
re-investigate CS4 and CS5 then.
4.2. Loop pairing and dynamic considerations
From the above analysis, it is better to fix organic
refluxratio, thus, the manipulated variables used for the
rectifiertray temperature at tray #5 and CSTR level loops shouldbe
picked from the below three manipulated variables:EtOH feed; AA
feed; or CSTR heat duty. Fig. 12 showsthe open-loop dynamic
response with organic reflux ratiofixed and step changes of either
EtOH feed, AA feed, orCSTR duty. When one of the three manipulated
variablesstep changes, the remaining two manipulated variablesare
kept at their original base case values. The first obser-vation is
that none of the dynamic responses can be mod-
-
Trays of 2nd Column1 2 3 4 5 6 7
Del
ta te
mpe
ratu
re (
C)
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6AA-5%
EtOH(H2O)+20%EtOH(H2O)-20%throughput+10%throughput-10%
Q1 fixed & AA fixed
Trays of 1ST Column1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
19 20
Del
ta te
mpe
ratu
re (
C)
-1.5
-1.0
-0.5
0.0
0.5
1.0
Fig. 10. Closed-loop sensitivity plot with CSTR duty and AA feed
fixed.
Trays of 2nd Column1 2 3 4 5 6 7
Del
ta te
mpe
ratu
re (
C)
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6AA-5%EtOH(H2O)+20%EtOH(H2O)-20%throughput+10%throughput-10%
Q1 fixed & EtOH fixed
Trays of 1ST Column1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
19 20
Del
ta te
mpe
ratu
re (
C)
-1.5
-1.0
-0.5
0.0
0.5
1.0
Fig. 11. Closed-loop sensitivity plot with CSTR duty and EtOH
feedfixed.
I-L. Chien et al. / Journal of Process Control 18 (2008) 215231
223eled as first-order plus deadtime process. A suitable modelto
use should be the integrating plus deadtime model form.Also from
this figure, one observed that the dynamicresponse of CSTR heat
duty versus tray #5 temperatureof rectifier is very problematic,
thus this pairing shouldbe avoid. The inverse response behavior can
be explainedby the combinatory effects of this tray temperature
dueto CSTR heat duty and also from the changing of theCSTR level
(reactor volume). It is also notice that theopen-loop effect of AA
feed to tray #5 temperature of rec-tifier is considerably smaller
than that from EtOH feed,thus, it is better to use EtOH feed to
control tray #5temperature.
Since CSTR heat duty is not suitable to use as themanipulated
variable for the tray #5 temperature controlloop, one of the
candidate overall control strategy is touse CSTR heat duty to
control CSTR level. This pairingwas also used in Chien et al. [1].
Fig. 13 shows anotherreason why AA feed is not suitable to control
tray #5temperature. With CSTR heat duty manipulating CSTRlevel in
automatic mode, we can do open-loop stepresponse for changes in AA
or EtOH feed flow rate.From Fig. 13, one observe that the open-loop
responsewith AA feed as the manipulated variable exhibits
largeinverse response. Let us take AA feed +1% changes asan
example. Because AA is a high-boiler with normalboiling point at
118 C, thus increasing AA feed intothe system should eventually
make tray #5 temperatureto increase. However, increasing AA feed
flow rate willalso result in increasing the CSTR level, thus
makingthe CSTR heat duty to drop which temporarily decreas-ing tray
#5 temperature causing the inverse response.Increasing EtOH feed
flow rate will not have the inverseresponse because EtOH is a
low-boiler thus eventuallywill make tray #5 temperature to drop
which coincidewith the effect of decreasing CSTR heat duty due to
levelcontrol action. Thus, the first overall control strategy
tofurther dynamically evaluate the control performance isto use
CSTR heat duty to control CSTR level and to
-
T 5 o
f 1st c
olum
n (
C)
94.4
94.6
94.8
95.0
95.2
95.4
95.6
Q1 +1%Q1 -1%
Time (hr)0 10 15 20 25 30
T 5 o
f 1st c
olum
n (
C)
92
94
96
98
100
ETOH feed +1%ETOH feed -1%
CST
R le
vel (
m)
4.356
4.358
4.360
4.362
4.364
4.366
4.368
4.370
4.372
ETOH feed +1%ETOH feed -1%
T 5 o
f 1st c
olum
n (
C)
94.2
94.4
94.6
94.8
95.0
95.2
95.4
95.6
95.8
AA feed +1%AA feed -1%
CST
R le
vel (
m)
4.358
4.360
4.362
4.364
4.366
4.368
4.370
AA feed +1% AA feed -1%
CST
R le
vel (
m)
4.355
4.360
4.365
4.370
Q1 +1%Q1 -1%
5Time (hr)
0 10 15 20 25 305Time (hr)
0 10 15 20 25 305
Time (hr)0 10 15 20 25 305
Time (hr)0 10 15 20 25 305
Time (hr)0 10 15 20 25 305
Fig. 12. Open-loop responses with step changes in either of EtOH
feed, AA feed, or CSTR duty.
224 I-L. Chien et al. / Journal of Process Control 18 (2008)
215231use EtOH feed flow rate to control tray #5 temperatureof
rectifier. AA feed will be used as the throughputmanipulator and
other control loops are explained previ-ously. This overall control
strategy is denoted as ControlStrategy (I). This overall control
strategy is similar towhat Chien et al. [1] used in their ethyl
acetate system.However, in their control strategy the tray
temperatureat rectifier was controlled by manipulating the acid
feedflow rate rather than the EtOH feed flow rate as is usedin our
system.
Another overall control strategy from previous CS3 is touse CSTR
heat duty as throughput manipulator and lettingorganic reflux ratio
to be fixed under various disturbances.Note again from previous
Fig. 9, this CS3 results in theleast temperature deviations under
perfect compositioncontrol. In this CS3, AA and EtOH feed flow
rates willbe used as the manipulated variables for the CSTR
levelloop and tray #5 temperature control loop at rectifier.The
loop pairing selection is to use AA feed to controlCSTR level and
to use EtOH feed to control tray #5 tem-perature at rectifier since
the effect of AA feed to this traytemperature is considerably
smaller as can be seen in previ-ous Fig. 12. This second overall
control strategy using twofeed flow rates to control CSTR level and
one tray temper-ature at rectifier is denoted as Control Strategy
(II). Kay-mak and Luyben [14] in their reactive distillation
columncontrol study proposed to use two feed flow rates to con-trol
two tray temperatures at the reactive distillation col-umn.
Although their control strategy is different than ourstudy, the
concept of using heat duty as throughput manip-ulator is the
same.4.3. Open-loop sensitivity analysis to verify the
suitability
of temperature control points
Before the dynamic evaluation of closed-loop perfor-mances for
the two overall control strategies, the open-loopsensitivity
analysis will be performed to verify that the tem-perature control
point determined by closed-loop sensitiv-ity analysis in Section
4.1 is really workable or not. Inorder to have the tray temperature
control loop to workwell, the chosen manipulated variable needs to
haveenough open-loop sensitivity to the controlled tray
temper-ature. For the open-loop sensitivity plots of Control
Strat-egy (I), CSTR level loop is in automatic mode manipulatedby
the CSTR heat duty. Similarly for the open-loop sensi-tivity plots
of Control Strategy (II), the CSTR level loop isin automatic mode
manipulated by AA feed flow rate. Sim-ilar open-loop sensitivities
are observed for either of Con-trol Strategies (I) or (II). Fig. 14
shows the exampleresult for Control Strategy (II). From this
figure, it isshown that enough open-loop sensitivity is observed
forthe selected temperature control points. The same conclu-sion
can be made for Control Strategy (I).
4.4. Closed-loop dynamic simulation results
Rigorous dynamic simulator, Aspen DynamicsTM, wasused for all
the closed-loop dynamic simulations. The moredifficult to handle
unmeasured feed composition changeswill be tested first on Control
Strategies (I) and (II).Fig. 15 shows the closed-loop dynamic
simulation withAA feed concentration changes from totally AA to
include
-
CSTR Level Q1 in auto mode
Time (hr)0 10 15 20 25 30
T 5of
1st C
olum
n (
C)
93.5
94.0
94.5
95.0
95.5
96.0
96.5
AA Feed +1%AA Feed -1%
CSTR Level Q1 in auto mode
T 5of
1st C
olum
n (
C)
92
93
94
95
96
97
98
99
EtOH Feed +1%EtOH Feed -1%
5
Time (hr)0 10 15 20 25 305
Fig. 13. Open-loop responses for tray #5 temperature of
rectifyingcolumn with step changes in either AA feed or EtOH feed
while CSTRlevel manipulating its duty is in auto mode.
Trays of 2nd Column1 2 3 4 5 6 7
Tem
pera
ture
(oC
)
75
80
85
90
95
100
105
Base caseQ2 +1%Q2 -1%
Trays of 1ST Column
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Tem
pera
ture
(oC
)
80
90
100
110
120
Base caseEtOH feed +0.01%EtOH feed -0.01%
Fig. 14. Open-loop sensitivity plot for the rectifying and
strippingcolumns while CSTR level is controlled by AA feed.
I-L. Chien et al. / Journal of Process Control 18 (2008) 215231
2255% water at t = 2 h. The three important control loops(tray #5
temperature at rectifier, tray #5 temperature atstripper, and CSTR
level) are all controlled back to theirsetpoint values. The rather
slowness of the dynamicresponse is mainly due to the needed large
volume of theCSTR reactor in the optimized flowsheet. With a
smallerCSTR reactor, the stringent AA impurity specification inthe
final product can not be satisfied. Since interactionsbetween the
CSTR level loop and rectifier T5 loop are quitesevere as can be
seen in previous Fig. 12, together with theunusual inverse response
characteristics in the off-diagonalelement (rectifier T5 vs. Q1),
the pattern of the closed-loopdynamic response is rather
unusual.
By looking at the final product composition, althoughthe EtOH
impurity in the final product stream holds nicelyusing this control
strategy, however, there is some deviationof the AA concentration
with this impurity changing from0.005 wt% to 0.0052 wt%. This
agrees with the closed-loopsensitivity plot in previous Fig. 7
where ideally the temper-ature setpoint at rectifier should be
adjusted a little in theface of AA feed composition
disturbance.Fig. 16 shows the simulation results for 20% changesin
the EtOH feed composition (water contents in this feedstream from
17.8 mol% to 21.35 mol% or from 17.8 mol%to 14.24 mol%) at t = 2 h.
The control strategy performswell for this type of disturbance with
both the AA andEtOH impurities in the product stream holding very
closeto their specifications.
The closed-loop responses of these two types of feedcomposition
changes for Control Strategy (II) are shownin Figs. 17 and 18. It
is noticed that for Control Strategy(II) both AA and EtOH
impurities are very close to theirspecifications despite various
feed composition changes.This agrees with the earlier findings via
steady-state simu-lation (see Fig. 9) where no adjustment of the
temperaturesetpoints are needed for AA or EtOH feed
compositionchanges.
It is also noticed that the transient response also favorControl
Strategy (II) in that the ranges of variations inthe two product
impurities during the transient periodare much narrower for Control
Strategy (II). For example,
-
Time (hr)0 50 100 150 200 250 300
Fin
al p
rodu
ct A
A C
once
ntra
tion
2.5e-5
3.0e-5
3.5e-5
4.0e-5
4.5e-5
5.0e-5
5.5e-5
6.0e-5
Time (hr)0 50 100 150 200 250 300
Fin
al p
rodu
ct E
tOH
Con
cent
ratio
n
9.900e-4
9.950e-4
1.000e-3
1.005e-3
1.010e-3
1.015e-3
Time (hr)0 50 100 150 200 250 300
T5
of 1
st C
olum
n (o
C)
92
93
94
95
96
Time (hr)0 50 100 150 200 250 300
EtO
H F
eed
Rat
e (m
ol/m
in)
94
96
98
100
102
104
Time (hr)0 50 100 150 200 250 300
T5
of 2
nd C
olum
n (o
C)
99.615
99.620
99.625
99.630
99.635
99.640
99.645
Time (hr)0 50 100 150 200 250 300
Reb
oile
r D
uty
of 2
nd C
olum
n (K
W)
60
62
64
66
68
70
72
Time (hr)0 50 100 150 200 250 300
CS
TR
Lev
el (
m)
4.360
4.361
4.362
4.363
4.364
4.365
Time (hr)0 50 100 150 200 250 300
Reb
oile
r D
uty
of 1
st C
olum
n (K
W)
140
142
144
146
148
150
152
154
AA Concentration-5%
5.20e-51.007e-3
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.
226 I-L. Chien et al. / Journal of Process Control 18 (2008)
215231with 20% changes in the EtOH feed composition, therange of
variations in AA impurity is from 0.0040 wt% to0.0063 wt% for
Control Strategy (II). However, much lar-ger 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 followsthe
same tuning rules. The CSTR level control loop istuned first with
IMC-PID tuning method of Chien andFruehauf [15] using integrating
plus deadtime model. Theclosed-loop time constant is set to be
twice of the modelapparent deadtime. For the Control Strategy (II),
sincethe CSTR level is controlled by the AA feed rate whichhave
direct influence on the level, the tuning is tighter thanthe result
of Control Strategy (I). However, by lookingat the dynamic response
of AA feed rate in Fig. 17,although the tuning is tight, the
manipulated variable isquite acceptable with small overshoot. After
this importantlevel control loop is put into automatic mode, the
temper-ature loop at stripper is tuned next followed by the
tuningof the temperature loop at rectifier. The dynamics of the
-
Time (hr)0 50 100 150 200 250 300
Fina
l pro
duct
AA
Con
cent
ratio
n
2e-5
3e-5
4e-5
5e-5
6e-5
7e-5
8e-5
9e-5
Time (hr)
Time (hr) Time (hr)
Time (hr) Time (hr)
Time (hr) Time (hr)
0 50 100 150 200 250 300
Fina
l pro
duct
EtO
H C
once
ntra
tion
9.850e-4
9.900e-4
9.950e-4
1.000e-3
1.005e-3
1.010e-3
1.015e-3
0 50 100 150 200 250 300
T 5 o
f 1st
Col
umn
(oC
)
91
92
93
94
95
96
97
98
99
0 50 100 150 200 250 300Et
OH
Fee
d R
ate
(mol
/min
)
94
96
98
100
102
104
106
108
110
0 50 100 150 200 250 300
T 5 o
f 2nd
Col
umn
(oC
)
99.59
99.60
99.61
99.62
99.63
99.64
99.65
0 50 100 150 200 250 300
Reb
oile
r Dut
y of
2nd
Col
umn
(KW
)
67
68
69
70
71
72
73
74
0 50 100 150 200 250 300
CST
R L
evel
(m)
4.360
4.362
4.364
4.366
4.368
4.370
0 50 100 150 200 250 300
Reb
oile
r Dut
y of
1st C
olum
n (K
W)
146
148
150
152
154
156
158
EtOH Feed H2O Concentration +20%EtOH Feed H2O Concentration
-20%
4.90e-5
1.002e-3
5.03e-5
9.974e-4
Fig. 16. Closed-loop performance for Control Strategy (I) with
EtOH feed concentration 20% changes at t = 2 h.
I-L. Chien et al. / Journal of Process Control 18 (2008) 215231
227temperature loop at stripper is much faster than the one
atrectifier, thus it is tuned first and put into automatic
modebefore the tuning of the final temperature loop at
rectifier.Integrating plus deadtime model was also used with
theclosed-loop time constant to be twice of the model appar-ent
deadtime. The tuning of this final temperature controlloop is
crucial for the overall control performance. Thesuitable
combination of the controller gain and reset timeis important to
achieve faster closed-loop response withoutmuch of the
oscillation.
Since Control Strategy (II) works better to handle theunmeasured
feed composition disturbances, this controlstrategy is further
tested with simulation study for the casewith throughput changes.
For this control strategy,increasing or decreasing the production
rate of the finalproduct needs to go through the changes of the
CSTR heat
-
0 50 100 150 200 250 300
Fina
l pro
duct
AA
Con
cent
ratio
n
4.84e-54.86e-54.88e-54.90e-54.92e-54.94e-54.96e-54.98e-55.00e-55.02e-55.04e-5
0 50 100 150 200 250 300
Fina
l pro
duct
EtO
H C
once
ntra
tion
9.980e-4
9.990e-4
1.000e-3
1.001e-3
1.002e-3
1.003e-3
1.004e-3
0 50 100 150 200 250 300
T 5 o
f 1st
Col
umn
(oC
)
94.85
94.90
94.95
95.00
95.05
95.10
0 50 100 150 200 250 300Et
OH
Fee
d R
ate
(mol
/min
)100.4
100.6
100.8
101.0
101.2
101.4
101.6
101.8
102.0
0 50 100 150 200 250 300
T 5 o
f 2nd
Col
umn
(oC
)
99.59
99.60
99.61
99.62
99.63
99.64
99.65
0 50 100 150 200 250 300
Reb
oile
r Dut
y of
2nd
Col
umn
(KW
)
67.0
67.5
68.0
68.5
69.0
69.5
70.0
70.5
0 50 100 150 200 250 300
CST
R L
evel
(m)
4.36405
4.36410
4.36415
4.36420
4.36425
4.36430
4.36435
0 50 100 150 200 250 300
AA fe
ed R
ate
(mol
/min
)
65.5
66.0
66.5
67.0
67.5
68.0
68.5
69.0
69.5
AA Concentration-5%
5.02e-5 1.003e-3
Time (hr) Time (hr)
Time (hr) Time (hr)
Time (hr) Time (hr)
Time (hr) Time (hr)
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.
228 I-L. Chien et al. / Journal of Process Control 18 (2008)
215231duty. Fig. 19 shows the simulation results with 10%changes of
the CSTR heat duty. With these changes ofthe CSTR heat duty, the
two feed flow rates are changedaccordingly with only small
deviations in the AA productimpurity. Because this disturbance is
considered as aknown load change, thus similar to the paper by
Huanget al. [13], a planning of the temperature setpoint
adjust-ment versus each CSTR heat duty can easily be made
fromprocess simulation to circumvent the impurity deviationproblem.
For example, with 10% changes in the through-put manipulator (CSTR
heat duty), the setpoint of tray #5temperature at rectifier should
decrease around 0.3 Caccording to Fig. 9 so that AA impurity in the
final productcan return back to 0.005 wt%.
With above 10% changes of the CSTR heat duty, theproduct rate is
changed from 63.8824 mol/min to69.66846 mol/min (a +9.1% increase
in production) orfrom 63.8824 mol/min to 58.0364 mol/min (a
9.1%
-
Time (hr)0 50 100 150 200 250 300
Fin
al p
rodu
ct A
A C
once
ntra
tion
3.5e-5
4.0e-5
4.5e-5
5.0e-5
5.5e-5
6.0e-5
6.5e-5
Time (hr)
Time (hr) Time (hr)
Time (hr) Time (hr)
Time (hr) Time (hr)
0 50 100 150 200 250 300
Fin
al p
rodu
ct E
tOH
Con
cent
ratio
n
9.850e-4
9.900e-4
9.950e-4
1.000e-3
1.005e-3
1.010e-3
1.015e-3
0 50 100 150 200 250 300
T5
of 1
st C
olum
n (o
C)
93.5
94.0
94.5
95.0
95.5
96.0
96.5
97.0
0 50 100 150 200 250 300
EtO
H F
eed
Rat
e (m
ol/m
in)
92
94
96
98
100
102
104
106
108
110
112
0 50 100 150 200 250 300
T5
of 2
nd C
olum
n (o
C)
99.600
99.605
99.610
99.615
99.620
99.625
99.630
99.635
99.640
0 50 100 150 200 250 300
Reb
oile
r D
uty
of 2
nd C
olum
n (K
W)
67
68
69
70
71
72
73
0 50 100 150 200 250 300
CS
TR
Lev
el (
m)
4.3640
4.3641
4.3642
4.3643
4.3644
4.3645
0 50 100 150 200 250 300
AA
Fee
d R
ate
(mol
/min
)
63
64
65
66
67
68
69
EtOH Feed H2O Concentration +20%
EtOH Feed H2O Concentration -20%
4.98e-5
4.95e-5
1.00e-3
9.96e-4
Fig. 18. Closed-loop performance for Control Strategy (II) with
EtOH feed concentration 20% changes at t = 2 h.
I-L. Chien et al. / Journal of Process Control 18 (2008) 215231
229decrease in production). This demonstrates that CSTRheat duty
can be used as a throughput manipulator tosmoothly increase or
decrease the production rate of thefinal product. The final
proposed overall control strategyfor this process can be seen in
Fig. 20.
5. Conclusions
In this paper, design and control of a complete ethylacrylate
process with coupled reactor/columns configura-tion has been
investigated. Unlike other paper in the liter-ature with only one
reactive distillation column toproduce products like methyl acetate
or butyl acetate, thisoverall process flowsheet is more complex
including aCSTR, a rectifying column, a decanter, and another
strip-ping column. The design procedure is based on the
max-imization of total annual profit (TAP) for the overallsystem.
The final EA product is having very high purityof 99.78 mol% (or
99.89 wt%) with two impurities ofAA and EtOH meeting stringent
specifications of
-
Time (hr)0 50 100 150 200 250 300
Fina
l pro
duct
AA
Con
cent
ratio
n
3.5e-5
4.0e-5
4.5e-5
5.0e-5
5.5e-5
6.0e-5
6.5e-5
7.0e-5
Time (hr)
Time (hr) Time (hr)
Time (hr) Time (hr)
Time (hr) Time (hr)
0 50 100 150 200 250 300
Fina
l pro
duct
EtO
H C
once
ntra
tion
9.20e-4
9.40e-4
9.60e-4
9.80e-4
1.00e-3
1.02e-3
1.04e-3
1.06e-3
1.08e-3
0 50 100 150 200 250 300
T 5 o
f 1st
Col
umn
(oC
)
93
94
95
96
97
98
0 50 100 150 200 250 300
EtO
H F
eed
Rat
e (m
ol/m
in)
85
90
95
100
105
110
115
120
0 50 100 150 200 250 300
T 5 o
f 2nd
Col
umn
(oC
)
99.50
99.55
99.60
99.65
99.70
99.75
0 50 100 150 200 250 300
Reb
oile
r Dut
y of
2nd
Col
umn
(KW
)
6062646668707274767880
0 50 100 150 200 250 300
CST
R L
evel
(m)
4.362
4.364
4.366
4.368
0 50 100 150 200 250 300
AA F
eed
Rat
e (m
ol/m
in)
5658606264666870727476
Throughput +10%Throughput-10%
4.77e-5
1.01e-35.24e-5
9.89e-4
Fig. 19. Closed-loop performance for Control Strategy (II) with
10% changes in CSTR duty at t = 2 h.
230 I-L. Chien et al. / Journal of Process Control 18 (2008)
2152310.005 wt% AA and 0.1 wt% EtOH in the final productstream.
For the overall control strategy of this process,
astraightforward procedure has been followed which usesclosed-loop
sensitivity analysis to screen out the worsecandidates of overall
control strategy and to pick thesuitable temperature control
points. The final recom-mended overall control strategy is found to
be differentthan another similar coupled reactor/columns process
pub-lished earlier [1]. Both AA and EtOH feed flow rates areused as
manipulated variables in the overall control strat-egy to control
CSTR level and one tray temperature at rec-tifier with CSTR heat
duty left as the throughputmanipulator for the overall process. The
final EA productwith stringent specifications of EtOH and AA
impuritiescan be achieved with this proposed overall control
strategydespite various feed composition disturbances andthroughput
changes.
-
Organic Reflux
AA
Steam
Aqueous
EtOH
Decanter
Steam
EA
Reboiler
Water
CSTR
PC PCLC LC
LC
LC
FC
FC
FC
FC
1 1
196
207
TCTC
X
X
TC
TC
(throughput manipulator)
Fig. 20. Proposed overall control strategy of this process.
I-L. Chien et al. / Journal of Process Control 18 (2008) 215231
231Acknowledgements
The first author, I-Lung Chien, take this opportunity tothank
Prof. Dale Seborg for his guidance and valuable ad-vices over the
years. This paper was prepared while I-Lungwas back visiting
Department of Chemical EngineeringUCSB after completing his Ph.D.
degree as Dales studentmore than 20 years ago. There is an old
Chinese sayingwhich conveys I-Lungs appreciation to Prof.
Seborg:Once becoming a teacher of yours, always respect himas 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 ofan ethyl acetate process: coupled reactor/column
configuration, J.Proc. Cont. 15 (2005) 435449.
[2] M. Witczak, W. Grzesik, J. Skrzypek, Kinetyka estryfikacji
kwasuakrylowego nizszymi alkoholami alifatycznymi, Inz. Chem.
Proces-owa 25 (2004) 331340.
[3] C.K. Yi, W.L. Luyben, Design and control of coupled
reactor/column systems Part 1. A binary coupled reactor/rectifier
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/columnsystem, 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 andrecycle,
Comput. Chem. Eng. 21 (1997) 6986.
[6] S.F. Chiang, C.L. Kuo, C.C. Yu, D.S.H. Wong, Design
alternativesfor the amyl acetate process: coupled reactor/column
and reactivedistillation, Ind. Eng. Chem. Res 41 (2002)
32333246.
[7] U. Weidlich, J. Gmehling, A modified UNIFAC model. 1.
Predictionof VLE, HE and c1, Ind. Eng. Chem. Res. 26 (1987)
13721381.
[8] J. Gmehling, J. Li, M. Schiller, A modified UNIFAC model.
2.Present parameter matrix and results for different
thermodynamicproperties, Ind. Eng. Chem. Res. 32 (1993) 178193.
[9] J.G. Hayden, J.P. OConnell, A generalized method for
predictingsecond virial coefficients, 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, Engle-wood Cliffs, NJ, 1998.
[12] H.Y. Lee, H.P. Huang, I-L. Chien, Control of reactive
distillationprocess for production of ethyl acetate, J. Proc. Cont.
17 (2007) 363377.
[13] S.G. Huang, C.L. Kuo, S.B. Huang, Y.W. Chen, C.C. Yu,
Temper-ature control of heterogeneous reactive distillation, AIChE
J. 50(2004) 22032216.
[14] D.B. Kaymak, W.L. Luyben, Comparison of two types of
two-temperature control structures for reactive distillation
columns, Ind.Eng. Chem. Res. 44 (2005) 45254640.
[15] I-L. Chien, P.S. Fruehauf, Consider IMC tuning to
improvecontroller performance, Chem. Eng. Prog. 86 (1990) 3341.
Overall control strategy of a coupled reactor/columns process
for the production of ethyl acrylateIntroductionThermodynamic and
kinetic model used in the simulationDesign flowsheet of the
complete processOverall control strategy of this processDetermine
of tray temperature control point and the control structureLoop
pairing and dynamic considerationsOpen-loop sensitivity analysis to
verify the suitabilityof temperature control pointsClosed-loop
dynamic simulation results
ConclusionsAcknowledgementsReferences