- Email: [email protected]
Dynamic controllability comparison of reactive distillation columns with single and double reactive sections for two-stage consecutive reactions
Accepted Manuscript Title: Dynamic Controllability Comparison of Reactive Distillation Columns with Single and Double Reactive Sections for Two-Stage Consecutive Reactions Authors: Ilayda N. Oksal, Devrim B. Kaymak PII: DOI: Reference:
S0263-8762(17)30643-3 https://doi.org/10.1016/j.cherd.2017.11.025 CHERD 2904
To appear in: Received date: Revised date: Accepted date:
3-7-2017 2-11-2017 17-11-2017
Please cite this article as: Oksal, Ilayda N., Kaymak, Devrim B., Dynamic Controllability Comparison of Reactive Distillation Columns with Single and Double Reactive Sections for Two-Stage Consecutive Reactions.Chemical Engineering Research and Design https://doi.org/10.1016/j.cherd.2017.11.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Paper submitted to Chemical Engineering Research and Design
IP T
Dynamic Controllability Comparison of Reactive Distillation Columns with Single and Double Reactive Sections for Two-Stage Consecutive Reactions
SC R
Ilayda N. Oksal
M
34469, Maslak, Istanbul, Turkey
N
Istanbul Technical University
A
Department of Chemical Engineering
U
Devrim B. Kaymak*
*
A
CC E
Highlights:
PT
ED
To whom correspondence should be addressed: [email protected]; Phone: +90-212-285-3539; Fax: +90-212-285-2925
Controllability of reactive columns including two consecutive reactions is studied
Columns with single (SRS) and double (DRS) reactive sections are compared
RDC-DRS is found as an acceptable alternative in terms of design and control
Abstract
The purpose of this study is to compare the dynamic controllability of different reactive distillation column configurations including two stage consecutive reversible reactions. Comparison is done between two configurations; the conventional column with single reactive section (RDC-SRS) and the novel column with double reactive sections (RDC-DRS) using two illustrative examples; one ideal generic system and one real system which is the
IP T
transesterification of dimethyl carbonate with ethanol to form diethyl carbonate. The dynamic
results demonstrate that the control performance of RDC-SRS and RDC-DRS are quite similar
SC R
for both case studies. Thus, the novel RDC-DRS is an acceptable alternative to the conventional RDC-SDR for the separation of two-stage consecutive reversible reactions in terms of both
N
U
steady-state and control performances.
A
1. Introduction
M
The increasing environmental and economic concerns enhance the need of process intensification in chemical industry. Reactive distillation, which combines reaction and
ED
separation in a single unit, stands out as one of the successful examples of process intensification in both the academia and the industry. Thus, the number of patents and research
CC E
decades.
PT
papers on design and control of the reactive distillation columns have rapidly increased for two
A literature search conducted by Luyben and Yu shows that there are over 235 reaction system studied in literature. Once these systems are classified according to their reaction types, almost
A
40% of the systems falls into two reactant-two product (A+B↔C+D) category, while more than 25% of them takes part in two reactant-one product (A+B↔C) form. Following them is the two-stage reactions with almost 15% in total [1].
Consecutive reversible reactions (A+B↔C+D and C+B↔E+D or A+B↔C and C+B↔D) are one of the common forms of two-stage reactions with important industrial applications in transesterification and esterification [2-7]. In all of these design studies, reactive distillation columns with single reactive section (RDC-SRS) have been investigated to separate two-stage
IP T
consecutive reaction systems.
Recently, Yu et al. compared the steady-state designs of RDC-SRS and reactive distillation
SC R
column with double reactive sections (RDC-DRS) to separate two-stage consecutive reaction
systems [8]. In the case of RDC-DRS, the total number of reactive trays are divided into two separate reactive sections by adding a middle non-reactive zone. The common reactant used in
U
both reactions is splitted and fed into both reactive sections, while the other reactant is only fed
N
into upper reactive section. The idea behind is that the first stage reaction takes place
A
dominantly in the upper reactive section, while the second stage reaction occurs principally in
M
the lower reactive section. Two different case-studies, one generic and one real, have been studied to compare the steady-state performances of reactive columns with single- and double
ED
reactive sections. Results of both case-studies illustrate that RDC-DRS has a lower energy cost
PT
compared to RDC-SRS, which makes RDC-DRS a possible alternative to conventional reactive columns in terms of steady-state design. On the other hand, in order to indicate the RDC-DRS
CC E
as a highly competitive alternative to RDC-SRS, their dynamic performances should be also investigated and compared.
A
However, there are only a handful of case studies examining the controllability of reactive distillation columns including two-stage consecutive reaction systems. Hung et al. studied the plantwide control of a process including three distillation columns, where the first column is a reactive column, for the esterification of adipic acid and methanol to produce dimethyl adipate [9]. Wei et al. proposed a control strategy for a two-column process for the production of diethyl
carbonat [10]. Consecutive reaction system with two irreversible reactions studied by Bo et al. was the benzene chloride production from toluene and chlorine, and a single reactive distillation column was used [11]. The common point of all these control studies is using RDC-SRS with control strategies including temperature controllers.
IP T
On the other hand, the results of a recent study on the controllability of a RDC-DRS for a generic two-stage consecutive reaction system shows that using only inferental temperature
SC R
controller displays poor performance, hence a workable control structure should include direct control of compositions [12]. However, neither any comparison with the controllability of RDC-SRS for the generic system is conducted nor the controllability of any real system as an
U
illustrative example is studied there.
N
The aim of this study is to compare the control performance and robustness of RDC-SRS and
A
RDC-DRS for the separation of two-stage consecutive reversible reactions using two
M
illustrative examples; an ideal generic reaction system and a real reaction system. The real
ED
system studied is the transesterification of dimethyl carbonate with ethanol.
PT
2. Steady-State Design
For both generic and real systems, kinetic and physical properties parameters are are taken from
CC E
Yu et al. [8]. The design specifications such as the number of trays, the feed locations, the feed splitting ratio of the common reactant and the operating pressures are also kept same as those of the paper written by Yu et al. [8].
A
2.1 Ideal System
The first system is a generic liquid-phase reaction
A B C D
(1)
CBED
(2)
with the net reaction rates for component j on tray i are given by r1 M k F ,1 x A x B k R,1 xC x D
(3)
r2 M k F , 2 xC x B k R, 2 x E x D
(4)
IP T
where concentrations are in mole fractions and liquid holdups are in kmoles. The forward and backward specific reaction rates follow the Arrhenius Law.
components
are
SC R
Ideal vapor-liquid equilibrium is assumed with constant volatilities. The volatilities of
D B C A E . Physicochemical properties and design
N
U
specifications are given in Table 1.
A
The flowsheets of RDC-SRS and RDC-DRS for the generic case are given in Figure 1. The
M
total number of trays is 28 (excluding the condenser and the reboiler) for both configurations. The RDC-SRS consists of stripping, rectifying and a single reactive sections. Reactive section
ED
is between tray 10 and tray 21 (stage numbering is bottom-up). The heavy reactant is fed from the top reactive tray, while the light reactant is introduced from the bottom reactive tray. The
PT
light product D leaves in the distillate, while the heavy product E is removed from the bottoms.
CC E
On the other hand, there are two reactive and one middle non-reactive sections in RDC-DRS besides the stripping and rectifyig sections. The total number of reactive trays is same with that of RDC-SRS, but they are separated into two sections with the middle non-reactive zone in
A
between. Although the heavy reactant A, which takes part only in the first reaction, is fed from the top of the upper reactive section, the light reactant B, which takes part in both reactions, is introduced from the bottom of both reactive sections.
The product purity specifications are 95 mol% D in the distillate and 95 mol% E in the bottoms. The optimum design of RDC-DRS scales down the reboiler heat duty by 4.5% comparing with RDC-SRS.
2.2 Real System
to produce diethyl carbonate (DEC) and methanol (MeOH).
SC R
DMC + EtOH EMC + MeOH
IP T
The second system is the transesterification of dimethyl carbonate (DMC) and ethanol (EtOH)
(6)
U
EMC + EtOH DEC + MeOH
(5)
N
The net reaction rates are given by
(7)
M
A
𝑟1 = 𝑘𝐹,1 ∗ 𝑥𝐷𝑀𝐶 ∗ 𝑥𝐸𝑇𝑂𝐻 − 𝑘𝑅,1 ∗ 𝑥𝐸𝑀𝐶 ∗ 𝑥𝑀𝐸𝑂𝐻
(8)
ED
𝑟2 = 𝑘𝐹,2 ∗ 𝑥𝐸𝑀𝐶 ∗ 𝑥𝐸𝑇𝑂𝐻 − 𝑘𝑅,2 ∗ 𝑥𝐷𝐸𝐶 ∗ 𝑥𝑀𝐸𝑂𝐻
Kinetic and physical properties and vapor-liquid equilibrium parameters are given in Table 2.
PT
The UNIQUAC-Redlich Kwong (UNIQUAC-RK) model is chosen to define the vapor-liquid equilibrium relationship of mixture. The binary parameters are obtained from the study of Wei
CC E
et al. [10].
The configurations of RDC-SRS and RDC-DRS for the real system are shown in Figure 2. The
A
RDC-SRS consists of stripping, rectifying and a single reactive sections. The total number of trays is 45 (including the condenser and the reboiler) for both configurations. For RDC-SRS, the reactive section runs from tray 4 to tray 39 (stage numbering is top to bottom). Dimethyl carbonate and ethanol are fed from the bottom and top trays of the reactive section, respectively. The heaviest component of the system is the main product DEC, while the lightest one is the
azeotropic mixture of DMC+MeOH, contrary to expectations, it is not the second product MeOH. Thus, an excess of EtOH is fed to the column to achieve a high conversion of DMC. The main product DEC is removed from the bottoms, while the distillate includes a mixture of MeOH and the excess EtOH. In the RDC-DRS, upper and lower reactive sections run from tray 2 to 30, and tray 34 to 40, respectively. Thus, the RDC-DRS does not have a rectifying section.
IP T
Dimethyl carbonate, which takes part only in the first reaction, is fed from the bottom of the upper reactive section. On the other hand, ethanol taking part in both reactions is introduced
SC R
from the top of the upper reactive section (FETOH1) and from the bottom of the lower reactive section (FETOH2).
U
The bottom composition of DEC is set at 99.5 mol% by adjusting the distillate flowrate, while
N
the reflux ratio is adjusted to keep the conversion of DMC at 0.995. The optimum design of
A
RDC-DRS scales down the reboiler heat duty by 20.33% comparing with RDC-SRS. The
M
temperature profiles of ideal and real systems are given in Figure 3. The solid lines are for the
ED
single section columns, while the dashed lines are for the double section columns.
3. Dynamic Control of Ideal System
PT
The dynamic model and the control structure design of reactive distillation columns are
CC E
developed in Matlab. The overall control structure of the ideal system is developed to hold the product purities of distillate and bottoms streams at their specifications against disturbances. Since the column pressure dynamics is much faster than the dynamics of other variables, it is
A
assumed that the pressure control is perfect. In all control structures, liquid levels are controlled by P-only controllers with a gain of 2. PI controllers are used for temperature and composition control loops. Two 60-s first-order measurement lags are added in temperature control loops, while, a 3 minutes dead time is included for the composition control loops. Ultimate gain (KU) and ultimate period (PU) of these controllers are obtained using ATV method. Controller
parameters such as KC and τI are calculated by Tyreus-Luyben settings. Temperature and composition transmitter spans are selected 50 K and 0.5, respectively. For inferential temperature control structures, the selection of trays to be controlled are done by using sensitivity analysis. All control valves are designed to be half open at steady state. Once the control structures are established, the processes are subjected to disturbances such as changes
IP T
in the production rate handle and the feed composition to analyze their effectiveness. 3.1 Control of RDC-SRS
SC R
First control structure studied is a two-point temperature control strategy, and shown in Figure
4. In this control structure, the fresh feed stream FA is flow controlled and acts as the production rate handle. Reflux drum level and column base level are controlled by manipulating the
U
distillate and bottoms flow rates, respectively. A ratio control keeps the vapor boilup in ratio
N
with bottoms flow rate. In accordance with fresh feed stream FB and reflux R are manipulated
A
to control T17 and T20, respectively. This control structure is labeled CS1I,S.
M
The closed loop response for ±20% changes in production rate handle are illustrated in Figure
ED
5. Result for the positive change is given by solid line, while dashed line shows the results for the negative change. The temperatures reach their steady-state values within five hours.
PT
Although stable closed-loop responses are obtained for this control structure, the product purities cannot be recovered, and they settle down into new steady-state values. Product purity
CC E
of the distillate stream deviates slightly from its setpoint, while there is a remarkable deviation in the product purity of the bottoms stream. Deviations in xB,E with a magnitude of 3.65 mol%
A
and 2.72 mol% for positive and negative 20% throughput changes, respectively, are unacceptable. Thus, the proposed two-point control structure fails to handle changes in disturbances. Second control structure given in Figure 6 includes one temperature and two composition loops. In this control structure, the fresh feed stream FA is used as the production rate handle. Distillate
and bottoms flow rates are manipulated to control the levels of reflux drum and column base, respectively. Stoichiometric balance is satisfied by manipulating fresh feed stream FB to control T17, while distillate and bottoms purities are controlled by manipulating reflux and vapor boilups, respectively. This control structure is labeled CS2I,S.
IP T
Figure 7A illustrates the closed loop response of this control structure for ±20% changes in production rate handle. This control structure gives a stable closed-loop response to the
SC R
disturbances tested. Although the transient deviation of the product purity in bottoms stream
rises up to 1 mol%, it is easily recovered to its setpoint in five hours. The product purity in distillate stream also settles down into its steady-state value in less than five hours with a much
N
balancing reaches its steady-state value in five hours.
U
smaller transient deviation. In addition, the temperature loop used for stoichiometric feed
A
Figure 7B demonstrates what happens when the composition zA of the fresh feed FA is changed
M
from pure A to a mixture including 5 mol% B. Although there is transient deviation in the
ED
product purity of bottoms stream up to 2 mol%, both product compositions and temperature of Tray 17 recover back to their set points easily. It is seen that the flow rate FB decreases to
PT
compensate the extra amount of B in feed stream FA.
CC E
3.2 Control of RDC-DRS
The three-point temperature control structure considered for RDC-DRS is given in Figure 8. In this structure, the fresh feed stream FA is flow controlled and serves as the production rate
A
handle. The reflux drum level is controlled by manipulating the distillate flow rate, while the bottoms flow rate is used to control the column base level. A ratio control keeps the bottoms in ratio with vapor boilup. Reflux and the fresh feed streams of light reactant (FB1 and FB2) are used to control the temperatures of three selected trays, which are Tray 21, Tray 13 and Tray 6, respectively. This control structure is labeled CS1I,D.
Figure 9 illustrates the closed loop responses for ±20% changes in production rate handle. Result for the positive change is given by solid line, while dashed line shows the results for the negative change. The temperatures reach their steady-state values within three hours. However, the product purities cannot be recovered, and they settle down into new steady-state values. Deviation of the product purity in the distillate stream is relatively small compared to the
IP T
remarkable deviation of the product purity in the bottoms stream. Deviations in xB,E with a magnitude of 1.95 mol% and 2.18 mol% for positive and negative 20% throughput changes,
SC R
respectively, are unacceptable. Thus, although the proposed control structure provides a stable control, it cannot properly handle changes in throughput.
Figure 10 shows the second control structure including composition and temperature
U
controllers. In this strategy, the compositions of product purities in distillate and bottoms
N
streams are directly controlled by manipulating the reflux and vapor boilup flowrates,
A
respectively. In addition, the temperatures on trays T13 and T6 are controlled by manipulating
M
the fresh feed streams FB1 and FB2, respectively, to satisfy the reactant stoichiometry
ED
inferentially. Other control loops are kept same with those of previous control structure. This control structure is labeled CS2I,D.
PT
Closed loop responses of this control structure for ±20% changes in production rate handle are given in Figure 11A. This control structure provides a stable closed-loop response for both
CC E
positive and negative throughput changes. Although the transient deviation of the product purity in bottoms stream rises up to 1 mol%, it is easily recovered to its setpoint in five hours. The
A
product purity in distillate stream also settles down into its steady-state value in less than five hours with a smaller transient deviation. Temperature loops used to satisfy the reactant stoichiometry also settle to their steady-state values in five hours with small transient deviations.
Figure 11B gives the response of CS2I,D to a feed impurity disturbance such as 5 mol% B in fresh feed stream FA. Both product compositions recover back to their set points easily. Transient deviations of product purities are smaller than 1 mol%. Decreasing flow rates of reactant B helps to control the temperatures and satisfy the reaction stoichiometry. Table 3 gives the controller pairings and tuning parameters of all control structures examined for the ideal
IP T
system.
SC R
4. Dynamic Control of Real System
Aspen Dynamics is used for the control structure design and pressure-driven dynamic simulation of the reactive distillation columns. The overall control structure of the real system
U
is developed to hold the bottoms product purity of the column at its specification against
N
disturbances. Since the top stream of the column includes a mixture of ethanol and methanol,
A
which could be separated using a second column, distillate product purity control is not aimed
M
in this study. Liquid levels are controlled by P-only controllers with a gain of 2. PI controllers
ED
including a single 60-s first-order measurement lag are used for temperature control loops. Controller tuning constants are calculated using ATV test and Tyreus–Luyben tuning rule
PT
embedded in the commercial software Aspen Dynamics. Temperature transmitter spans are selected 50 K, and all control valves are designed to be half open at steady state. Temperature
CC E
control trays are determined by performing open-loop sensitivity analysis. Then, disturbances such as changes in the production rate handle and the feed composition are applied to demonstrate their robustness and performance.
A
4.1 Control of RDC-SRS Schematic of the two-point temperature control structure proposed for RDC-SRS is illustrated in Figure 12. In this control structure, feed flowrate of ethanol is flow controlled and serves as the production rate handle. Reflux drum level and column base level are controlled by manipulating the distillate and bottoms flow rates, respectively. A ratio control keeps the reflux
flowrate in ratio with distillate flow rate. The condenser pressure is controlled by manipulating the condenser heat removal. According to the results of sensitivity analysis, tray 26 and tray 35 are selected the most sensitive locations for temperature control. Thus, reboiler heat duty and feed flowrate of dimethyl carbonate are arranged to control T26 and T35, respectively. This
IP T
control structure is labeled CS1R,S. Figure 13A gives the responses of CS1R,S to ±20% changes in production rate handle, FEtOH.
SC R
The solid lines illustrate the case of +20% change and the dashed lines illustrate the case of -
20% change. Results show that these disturbances can be easily handled with some oscillation. Both tray temperatures which are directly controlled settle down to their set points in 5 hours
U
by properly arranging the manipulated variables. In addition, the purity of diethyl carbonate,
N
which is the main product of the process, settles down to a new steady-state with fairly small
A
transient and steady-state deviations. The magnitude of the final steady-state deviation is
M
smaller than 0.2% for both disturbances in positive and negative directions.
ED
The response of CS1R,S to a 10% change in feed composition (XEtOH) is given in Figure 13B. Tray temperatures return to their set points in less than 4 hours. It also takes about 4 hours for
PT
diethyl carbonate purity to settle down to new steady-state with a quite small deviation. The
CC E
size of the deviation is less than 0.04%.
4.2 Control of RDC-DRS Figure 14 gives the control structure for RDC-DRS. The feed flowrate of ethanol in the upper
A
reactive section (FETOH1) is flow controlled and serves as production rate handle. The distillate and bottoms flow rates are manipulated to control the reflux drum level and column base level, respectively. A ratio control is used to keep the reflux flowrate in ratio with distillate flow rate. Condenser heat removal is manipulated to control the condenser pressure. As the result of the sensitivity analysis, T27 and T35 are controlled by manipulating the reboiler heat duty and feed
flowrate of dimethyl carbonate, respectively. In addition, a fixed ratio controller is added to the structure to satisfy the feed balance between FETOH1 and FETOH2 . This control structure is labeled CS1R,D. Figure 15A gives the responses of CS1R,D to changes in production rate handle, FETOH1. The
IP T
tray temperatures recovery back to their set points in around 2 hours. Although stable closedloop responses are obtained for this control structure, the product purity of diethyl carbonate in
SC R
the bottoms stream deviates slightly from its setpoint. Absolute steady-state deviation of diethyl
carbonate purity is less than 0.2% in the face of positive and negative disturbances. Figure 15B illustrates the responses of the CS1R,D to a 10% change in feed composition. It takes 3 hours for
U
the system to settle down to steady-state. Tray temperatures return to their set points, while the
N
change in diethyl carbonate purity is less than 0.01%. Table 4 gives the controller pairings and
M
A
tuning parameters of both control structures studied for the real system.
5. Conclusions
ED
In this paper, controllability of two reactive distillation column configurations including two stage consecutive reversible reactions is compared. First configuration is a conventional
PT
reactive distillation column with a single reactive section, while the second one is a novel
CC E
column with double reactive sections. Their control performances are evaluated based on two representative examples, a generic system and a real system for diethyl carbonate production. Since the neat reactive distillation columns act like pure integrators with respect to the reactants,
A
satisfying the desired conversion of reactants in the reactive zone is important, so they do not leave the system as impurity. For this reason, some tray temperatures in the reactive section should be effectively controlled to satisfy the necessary conversion. For that to happen, the temperature profile should be proper to inferentially control the temperature, which means there should be a notable temperature difference between adjacent trays. In this study, the
temperature difference between the trays of reactive section for the ideal system is much smaller than that of the real system. Thus, the results of generic system for both RDC-SRS and RDCDRS indicate that control structures including direct composition control besides the inferential temperature controllers are necessary to hold the product pruties at their specifications. On the other hand, using a control structure including only inferential temperature controllers is found
IP T
enough to provide an effective control for both RDC-SRS and RDC-DRS of the real system studied.
SC R
The steady-state deviations of product purities are given in Table 5, and these results
demonstrate that control performances of RDC-DRS are just as good as RDC-SRS for both reaction system. Thus, it can be concluded that the RDC-DRS can be an alternative to the
A
CC E
PT
ED
M
N
A
regard to steady-state and control performances.
U
conventional RDC-SRS for the separation of two-stage consecutive reversible reactions with
References (1) Luyben, W. L.; Yu, C. C. Reactive Distillation Design and Control, Wiley: New Jersey, 2008. (2) Ung, S; Doherty, M.F. Synthesis of reactive distillation systems with multiple equilibrium
IP T
chemical reactions, Industrial & Engineering Chemistry Research, 1995, 34, 2555-2565. (3) Luo, H. P.; Xiao, W.D. A reactive distillation process for a cascade and azeotropic reaction
SC R
system: Carbonylation of ethanol with dimethyl carbonate, Chemical Engineering Science, 2001, 56, 403-410.
(4) Hung, S. B.; Lai, I.; Huang, H. P.; Lee, M. J.; Yu C.C. Reactive distillation for two-stage
N
Chemistry Research, 2008, 47, 3076-3087.
U
reaction systems: Adipic acide and glutaric acid esterifications, Industrial & Engineering
A
(5) Orjuela, A.; Kolah, A.; Lira, A. T.; Miller, D. J. Mixed succinic Acid/acetic acid
ED
Research, 2011, 50, 9209-9220.
M
esterification with ethanol by reactive distillation, Industrial & Engineering Chemistry
(6) Keller, T.; Holtbruegge, J.; Gorak, A. Transesterification of dimethyl carbonate with
PT
ethanol in a pilot-scale reactive distillation column, Chemical Engineering Journal, 2012, 180, 309-322.
CC E
(7) Yao, X.; Huang, K.; Chen, H.; Li, S. Employing top−bottom recycled reactive distillation to the separations of adipic acid and glutaric acid esterifications, Industrial & Engineering Chemistry Research, 2013, 52, 16870-16879.
A
(8) Yu, C.; Yao, X., Huang, K.; Zhang, L.; Wang, S.; Chen, H. A reactive distillation column with double reactive sections for the separations of two-stage consecutive reversible reactions, Chemical Engineering and Processing, 2014, 79, 56-68.
(9) Hung, S. B.; Chen, J. H.; Lin, Y. D.; Huang, H. P.; Lee, M. J.; Ward, J. D.; Yu, C. C. Control of plantwide reactive distillation processes: Hydrolysis, transesterification and two-stage esterification, Journal of the Taiwan Instute of Chemical Engineers, 2010, 41, 382-402. (10)
Wei, H. Y.; Rokhmah, A.; Handogo, R.; Chien, I. Design and control of reactive
distillation process for the production of diethyl carbonate via two consecutive
(11)
IP T
transesterification reactions, Journal of Process Control, 2011, 21, 1193-1207.
Bo, C.; Tang, J.; Cui, M.; Feng, K.; Qiao, X.; Gao, F. Control and profile setting of
Engineering Chemistry Research, 2013, 52, 17465-17474. (12)
SC R
reactive distillation column for benzene chloride consecutive reaction system, Industrial &
Kaymak D. B.; Unlu, H.; Ofkeli, T. Control of a reactive distillation column with double
U
reactive sections for two-stage consecutive reactions, Chemical Engineering and
A
CC E
PT
ED
M
A
N
Processing, 2017, 113, SI, 86-93.
Figure Caption
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 1 – Flowsheets of reactive columns for ideal system: (A) RDC-SRS, (B) RDC-DRS
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 2 – Flowsheets of reactive columns for real system: (A) RDC-SRS, (B) RDC-DRS
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 3 – Temperature profiles: (A) Ideal system, (B) Real system
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 4 – Control structure of RDC-SRS for ideal system: CS1I,S
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 5 – Results of CS1I,S: ±20% FA
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 6 – Control structure of RDC-SRS for ideal system: CS2I,S
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 7 – Results of CS2I,S: (A) ±20% FA, (B) zA,B = 0.05
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 8 – Control structure of RDC-DRS for ideal system: CS1I,D
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 9 – Results of CS1I,D: ±20% FA
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 10 – Control structure of RDC-DRS for ideal system: CS2I,D
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 11 – Results of CS2I,D: (A) ±20% FA, (B) zA,B = 0.05
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 12 – Control structure of RDC-SRS for real system: CS1R,S
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 13 – Results of CS1R,S: (A) ±20% FEtOH, (B) zEtOH,DMC = 0.10
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 14 – Control structure of RDC-DRS for real system: CS1R,D
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 15 – Results of CS1R,D: (A) ±20% FEtOH1, (B) zEtOH1,DMC = 0.10
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Table 1. Kinetic and physical properties for the ideal example Parameter Value Column pressure (bar) 10 First stage activation energy (kcal mol-1) forward 30 reverse 30 -1 Second stage activation energy (kcal mol ) forward 30 reverse 30 First stage specific reaction rate at 366 K (kmol s-1 kmol-1) forward 0.008 reverse 0.004 Second stage specific reaction rate at 366 K (kmol s-1 kmol-1) forward 0.0008 reverse 0.0016 First stage chemical equilibrium constant at 366 K 2 Second stage chemical equilibrium constant at 366 K 0.5 -1 Heat of reaction (kcal mol ) 0 Heat of vaporization (kcal mol-1) 6.944 a Vapor pressure constants (AVP,j - BVP,j) A 10.96 – 3862 B 12.35 – 3862 C 11.65 – 3862 D 13.04 – 3862 E 10.27 – 3862 a ln P s A j VP , j BVP , j T with temperature in K and vapor pressure in bar
Value 1.013 9.553 9.553
0.0692 0.0346 0.0134 0.0298 2 0.45 0
IP T
8.904 8.904
A
CC E
PT
ED
M
A
N
U
SC R
Table 2. Kinetic and physical properties for the real example Parameter Column pressure (bar) First stage activation energy (kcal mol-1) forward reverse Second stage activation energy (kcal mol-1) forward reverse First stage specific reaction rate at 366 K (kmol s-1 kmol-1) forward reverse Second stage specific reaction rate at 366 K (kmol s-1 kmol-1) forward reverse First stage chemical equilibrium constant at 366 K Second stage chemical equilibrium constant at 366 K Heat of reaction (kcal mol-1)
A
CC E
PT
ED
M
A
N
U
SC R
τI (min) 62.33 17.60 62.33 32.27 9.99 11.64 14.39 9.99 11.64 28.88
IP T
Table 3. Controller tuning parameters for ideal system Control structure CV MV KC T17 FB 0.50 CS1I,S T20 R 0.56 T17 FB 0.50 CS2I,S xD,D R 4.38 T6 FB2 0.84 CS1I,D T13 FB1 6.57 T21 R 0.70 T6 FB2 0.84 CS2I,D T13 FB1 6.57 xD,D R 5.52
Table 4. Controller tuning parameters for real system Control structure CV MV KC T26 QR 0.41 CS1R,S T35 FDMC 6.70 T27 QR 0.42 CS1R,D T35 FDMC 13.31
A
CC E
PT
ED
M
A
N
U
SC R
IP T
τI (min) 10.56 11.62 12.27 11.75
Table 5- Steady-state deviations Ideal System CS2I,S
CS2I,D |∆XB,E| / |∆XD,D|
+20% in PRH
0.00000 / 0.00000
0.00000 / 0.00000
-20% in PRH
0.00000 / 0.00000
0.00003 / 0.00001
5% impurity in PRH
0.00003 / 0.00001
0.00000 / 0.00000
CS1R,S
CS1R,D
|∆XDEC|
|∆XDEC|
+20% in PRH
0.00155
0.00192
-20% in PRH
0.00106
0.00123
10% impurity in PRH
0.00034
0.00001
A
CC E
PT
ED
M
A
N
SC R
U
Real System
IP T
|∆XB,E| / |∆XD,D|
S0263-8762(17)30643-3 https://doi.org/10.1016/j.cherd.2017.11.025 CHERD 2904
To appear in: Received date: Revised date: Accepted date:
3-7-2017 2-11-2017 17-11-2017
Please cite this article as: Oksal, Ilayda N., Kaymak, Devrim B., Dynamic Controllability Comparison of Reactive Distillation Columns with Single and Double Reactive Sections for Two-Stage Consecutive Reactions.Chemical Engineering Research and Design https://doi.org/10.1016/j.cherd.2017.11.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Paper submitted to Chemical Engineering Research and Design
IP T
Dynamic Controllability Comparison of Reactive Distillation Columns with Single and Double Reactive Sections for Two-Stage Consecutive Reactions
SC R
Ilayda N. Oksal
M
34469, Maslak, Istanbul, Turkey
N
Istanbul Technical University
A
Department of Chemical Engineering
U
Devrim B. Kaymak*
*
A
CC E
Highlights:
PT
ED
To whom correspondence should be addressed: [email protected]; Phone: +90-212-285-3539; Fax: +90-212-285-2925
Controllability of reactive columns including two consecutive reactions is studied
Columns with single (SRS) and double (DRS) reactive sections are compared
RDC-DRS is found as an acceptable alternative in terms of design and control
Abstract
The purpose of this study is to compare the dynamic controllability of different reactive distillation column configurations including two stage consecutive reversible reactions. Comparison is done between two configurations; the conventional column with single reactive section (RDC-SRS) and the novel column with double reactive sections (RDC-DRS) using two illustrative examples; one ideal generic system and one real system which is the
IP T
transesterification of dimethyl carbonate with ethanol to form diethyl carbonate. The dynamic
results demonstrate that the control performance of RDC-SRS and RDC-DRS are quite similar
SC R
for both case studies. Thus, the novel RDC-DRS is an acceptable alternative to the conventional RDC-SDR for the separation of two-stage consecutive reversible reactions in terms of both
N
U
steady-state and control performances.
A
1. Introduction
M
The increasing environmental and economic concerns enhance the need of process intensification in chemical industry. Reactive distillation, which combines reaction and
ED
separation in a single unit, stands out as one of the successful examples of process intensification in both the academia and the industry. Thus, the number of patents and research
CC E
decades.
PT
papers on design and control of the reactive distillation columns have rapidly increased for two
A literature search conducted by Luyben and Yu shows that there are over 235 reaction system studied in literature. Once these systems are classified according to their reaction types, almost
A
40% of the systems falls into two reactant-two product (A+B↔C+D) category, while more than 25% of them takes part in two reactant-one product (A+B↔C) form. Following them is the two-stage reactions with almost 15% in total [1].
Consecutive reversible reactions (A+B↔C+D and C+B↔E+D or A+B↔C and C+B↔D) are one of the common forms of two-stage reactions with important industrial applications in transesterification and esterification [2-7]. In all of these design studies, reactive distillation columns with single reactive section (RDC-SRS) have been investigated to separate two-stage
IP T
consecutive reaction systems.
Recently, Yu et al. compared the steady-state designs of RDC-SRS and reactive distillation
SC R
column with double reactive sections (RDC-DRS) to separate two-stage consecutive reaction
systems [8]. In the case of RDC-DRS, the total number of reactive trays are divided into two separate reactive sections by adding a middle non-reactive zone. The common reactant used in
U
both reactions is splitted and fed into both reactive sections, while the other reactant is only fed
N
into upper reactive section. The idea behind is that the first stage reaction takes place
A
dominantly in the upper reactive section, while the second stage reaction occurs principally in
M
the lower reactive section. Two different case-studies, one generic and one real, have been studied to compare the steady-state performances of reactive columns with single- and double
ED
reactive sections. Results of both case-studies illustrate that RDC-DRS has a lower energy cost
PT
compared to RDC-SRS, which makes RDC-DRS a possible alternative to conventional reactive columns in terms of steady-state design. On the other hand, in order to indicate the RDC-DRS
CC E
as a highly competitive alternative to RDC-SRS, their dynamic performances should be also investigated and compared.
A
However, there are only a handful of case studies examining the controllability of reactive distillation columns including two-stage consecutive reaction systems. Hung et al. studied the plantwide control of a process including three distillation columns, where the first column is a reactive column, for the esterification of adipic acid and methanol to produce dimethyl adipate [9]. Wei et al. proposed a control strategy for a two-column process for the production of diethyl
carbonat [10]. Consecutive reaction system with two irreversible reactions studied by Bo et al. was the benzene chloride production from toluene and chlorine, and a single reactive distillation column was used [11]. The common point of all these control studies is using RDC-SRS with control strategies including temperature controllers.
IP T
On the other hand, the results of a recent study on the controllability of a RDC-DRS for a generic two-stage consecutive reaction system shows that using only inferental temperature
SC R
controller displays poor performance, hence a workable control structure should include direct control of compositions [12]. However, neither any comparison with the controllability of RDC-SRS for the generic system is conducted nor the controllability of any real system as an
U
illustrative example is studied there.
N
The aim of this study is to compare the control performance and robustness of RDC-SRS and
A
RDC-DRS for the separation of two-stage consecutive reversible reactions using two
M
illustrative examples; an ideal generic reaction system and a real reaction system. The real
ED
system studied is the transesterification of dimethyl carbonate with ethanol.
PT
2. Steady-State Design
For both generic and real systems, kinetic and physical properties parameters are are taken from
CC E
Yu et al. [8]. The design specifications such as the number of trays, the feed locations, the feed splitting ratio of the common reactant and the operating pressures are also kept same as those of the paper written by Yu et al. [8].
A
2.1 Ideal System
The first system is a generic liquid-phase reaction
A B C D
(1)
CBED
(2)
with the net reaction rates for component j on tray i are given by r1 M k F ,1 x A x B k R,1 xC x D
(3)
r2 M k F , 2 xC x B k R, 2 x E x D
(4)
IP T
where concentrations are in mole fractions and liquid holdups are in kmoles. The forward and backward specific reaction rates follow the Arrhenius Law.
components
are
SC R
Ideal vapor-liquid equilibrium is assumed with constant volatilities. The volatilities of
D B C A E . Physicochemical properties and design
N
U
specifications are given in Table 1.
A
The flowsheets of RDC-SRS and RDC-DRS for the generic case are given in Figure 1. The
M
total number of trays is 28 (excluding the condenser and the reboiler) for both configurations. The RDC-SRS consists of stripping, rectifying and a single reactive sections. Reactive section
ED
is between tray 10 and tray 21 (stage numbering is bottom-up). The heavy reactant is fed from the top reactive tray, while the light reactant is introduced from the bottom reactive tray. The
PT
light product D leaves in the distillate, while the heavy product E is removed from the bottoms.
CC E
On the other hand, there are two reactive and one middle non-reactive sections in RDC-DRS besides the stripping and rectifyig sections. The total number of reactive trays is same with that of RDC-SRS, but they are separated into two sections with the middle non-reactive zone in
A
between. Although the heavy reactant A, which takes part only in the first reaction, is fed from the top of the upper reactive section, the light reactant B, which takes part in both reactions, is introduced from the bottom of both reactive sections.
The product purity specifications are 95 mol% D in the distillate and 95 mol% E in the bottoms. The optimum design of RDC-DRS scales down the reboiler heat duty by 4.5% comparing with RDC-SRS.
2.2 Real System
to produce diethyl carbonate (DEC) and methanol (MeOH).
SC R
DMC + EtOH EMC + MeOH
IP T
The second system is the transesterification of dimethyl carbonate (DMC) and ethanol (EtOH)
(6)
U
EMC + EtOH DEC + MeOH
(5)
N
The net reaction rates are given by
(7)
M
A
𝑟1 = 𝑘𝐹,1 ∗ 𝑥𝐷𝑀𝐶 ∗ 𝑥𝐸𝑇𝑂𝐻 − 𝑘𝑅,1 ∗ 𝑥𝐸𝑀𝐶 ∗ 𝑥𝑀𝐸𝑂𝐻
(8)
ED
𝑟2 = 𝑘𝐹,2 ∗ 𝑥𝐸𝑀𝐶 ∗ 𝑥𝐸𝑇𝑂𝐻 − 𝑘𝑅,2 ∗ 𝑥𝐷𝐸𝐶 ∗ 𝑥𝑀𝐸𝑂𝐻
Kinetic and physical properties and vapor-liquid equilibrium parameters are given in Table 2.
PT
The UNIQUAC-Redlich Kwong (UNIQUAC-RK) model is chosen to define the vapor-liquid equilibrium relationship of mixture. The binary parameters are obtained from the study of Wei
CC E
et al. [10].
The configurations of RDC-SRS and RDC-DRS for the real system are shown in Figure 2. The
A
RDC-SRS consists of stripping, rectifying and a single reactive sections. The total number of trays is 45 (including the condenser and the reboiler) for both configurations. For RDC-SRS, the reactive section runs from tray 4 to tray 39 (stage numbering is top to bottom). Dimethyl carbonate and ethanol are fed from the bottom and top trays of the reactive section, respectively. The heaviest component of the system is the main product DEC, while the lightest one is the
azeotropic mixture of DMC+MeOH, contrary to expectations, it is not the second product MeOH. Thus, an excess of EtOH is fed to the column to achieve a high conversion of DMC. The main product DEC is removed from the bottoms, while the distillate includes a mixture of MeOH and the excess EtOH. In the RDC-DRS, upper and lower reactive sections run from tray 2 to 30, and tray 34 to 40, respectively. Thus, the RDC-DRS does not have a rectifying section.
IP T
Dimethyl carbonate, which takes part only in the first reaction, is fed from the bottom of the upper reactive section. On the other hand, ethanol taking part in both reactions is introduced
SC R
from the top of the upper reactive section (FETOH1) and from the bottom of the lower reactive section (FETOH2).
U
The bottom composition of DEC is set at 99.5 mol% by adjusting the distillate flowrate, while
N
the reflux ratio is adjusted to keep the conversion of DMC at 0.995. The optimum design of
A
RDC-DRS scales down the reboiler heat duty by 20.33% comparing with RDC-SRS. The
M
temperature profiles of ideal and real systems are given in Figure 3. The solid lines are for the
ED
single section columns, while the dashed lines are for the double section columns.
3. Dynamic Control of Ideal System
PT
The dynamic model and the control structure design of reactive distillation columns are
CC E
developed in Matlab. The overall control structure of the ideal system is developed to hold the product purities of distillate and bottoms streams at their specifications against disturbances. Since the column pressure dynamics is much faster than the dynamics of other variables, it is
A
assumed that the pressure control is perfect. In all control structures, liquid levels are controlled by P-only controllers with a gain of 2. PI controllers are used for temperature and composition control loops. Two 60-s first-order measurement lags are added in temperature control loops, while, a 3 minutes dead time is included for the composition control loops. Ultimate gain (KU) and ultimate period (PU) of these controllers are obtained using ATV method. Controller
parameters such as KC and τI are calculated by Tyreus-Luyben settings. Temperature and composition transmitter spans are selected 50 K and 0.5, respectively. For inferential temperature control structures, the selection of trays to be controlled are done by using sensitivity analysis. All control valves are designed to be half open at steady state. Once the control structures are established, the processes are subjected to disturbances such as changes
IP T
in the production rate handle and the feed composition to analyze their effectiveness. 3.1 Control of RDC-SRS
SC R
First control structure studied is a two-point temperature control strategy, and shown in Figure
4. In this control structure, the fresh feed stream FA is flow controlled and acts as the production rate handle. Reflux drum level and column base level are controlled by manipulating the
U
distillate and bottoms flow rates, respectively. A ratio control keeps the vapor boilup in ratio
N
with bottoms flow rate. In accordance with fresh feed stream FB and reflux R are manipulated
A
to control T17 and T20, respectively. This control structure is labeled CS1I,S.
M
The closed loop response for ±20% changes in production rate handle are illustrated in Figure
ED
5. Result for the positive change is given by solid line, while dashed line shows the results for the negative change. The temperatures reach their steady-state values within five hours.
PT
Although stable closed-loop responses are obtained for this control structure, the product purities cannot be recovered, and they settle down into new steady-state values. Product purity
CC E
of the distillate stream deviates slightly from its setpoint, while there is a remarkable deviation in the product purity of the bottoms stream. Deviations in xB,E with a magnitude of 3.65 mol%
A
and 2.72 mol% for positive and negative 20% throughput changes, respectively, are unacceptable. Thus, the proposed two-point control structure fails to handle changes in disturbances. Second control structure given in Figure 6 includes one temperature and two composition loops. In this control structure, the fresh feed stream FA is used as the production rate handle. Distillate
and bottoms flow rates are manipulated to control the levels of reflux drum and column base, respectively. Stoichiometric balance is satisfied by manipulating fresh feed stream FB to control T17, while distillate and bottoms purities are controlled by manipulating reflux and vapor boilups, respectively. This control structure is labeled CS2I,S.
IP T
Figure 7A illustrates the closed loop response of this control structure for ±20% changes in production rate handle. This control structure gives a stable closed-loop response to the
SC R
disturbances tested. Although the transient deviation of the product purity in bottoms stream
rises up to 1 mol%, it is easily recovered to its setpoint in five hours. The product purity in distillate stream also settles down into its steady-state value in less than five hours with a much
N
balancing reaches its steady-state value in five hours.
U
smaller transient deviation. In addition, the temperature loop used for stoichiometric feed
A
Figure 7B demonstrates what happens when the composition zA of the fresh feed FA is changed
M
from pure A to a mixture including 5 mol% B. Although there is transient deviation in the
ED
product purity of bottoms stream up to 2 mol%, both product compositions and temperature of Tray 17 recover back to their set points easily. It is seen that the flow rate FB decreases to
PT
compensate the extra amount of B in feed stream FA.
CC E
3.2 Control of RDC-DRS
The three-point temperature control structure considered for RDC-DRS is given in Figure 8. In this structure, the fresh feed stream FA is flow controlled and serves as the production rate
A
handle. The reflux drum level is controlled by manipulating the distillate flow rate, while the bottoms flow rate is used to control the column base level. A ratio control keeps the bottoms in ratio with vapor boilup. Reflux and the fresh feed streams of light reactant (FB1 and FB2) are used to control the temperatures of three selected trays, which are Tray 21, Tray 13 and Tray 6, respectively. This control structure is labeled CS1I,D.
Figure 9 illustrates the closed loop responses for ±20% changes in production rate handle. Result for the positive change is given by solid line, while dashed line shows the results for the negative change. The temperatures reach their steady-state values within three hours. However, the product purities cannot be recovered, and they settle down into new steady-state values. Deviation of the product purity in the distillate stream is relatively small compared to the
IP T
remarkable deviation of the product purity in the bottoms stream. Deviations in xB,E with a magnitude of 1.95 mol% and 2.18 mol% for positive and negative 20% throughput changes,
SC R
respectively, are unacceptable. Thus, although the proposed control structure provides a stable control, it cannot properly handle changes in throughput.
Figure 10 shows the second control structure including composition and temperature
U
controllers. In this strategy, the compositions of product purities in distillate and bottoms
N
streams are directly controlled by manipulating the reflux and vapor boilup flowrates,
A
respectively. In addition, the temperatures on trays T13 and T6 are controlled by manipulating
M
the fresh feed streams FB1 and FB2, respectively, to satisfy the reactant stoichiometry
ED
inferentially. Other control loops are kept same with those of previous control structure. This control structure is labeled CS2I,D.
PT
Closed loop responses of this control structure for ±20% changes in production rate handle are given in Figure 11A. This control structure provides a stable closed-loop response for both
CC E
positive and negative throughput changes. Although the transient deviation of the product purity in bottoms stream rises up to 1 mol%, it is easily recovered to its setpoint in five hours. The
A
product purity in distillate stream also settles down into its steady-state value in less than five hours with a smaller transient deviation. Temperature loops used to satisfy the reactant stoichiometry also settle to their steady-state values in five hours with small transient deviations.
Figure 11B gives the response of CS2I,D to a feed impurity disturbance such as 5 mol% B in fresh feed stream FA. Both product compositions recover back to their set points easily. Transient deviations of product purities are smaller than 1 mol%. Decreasing flow rates of reactant B helps to control the temperatures and satisfy the reaction stoichiometry. Table 3 gives the controller pairings and tuning parameters of all control structures examined for the ideal
IP T
system.
SC R
4. Dynamic Control of Real System
Aspen Dynamics is used for the control structure design and pressure-driven dynamic simulation of the reactive distillation columns. The overall control structure of the real system
U
is developed to hold the bottoms product purity of the column at its specification against
N
disturbances. Since the top stream of the column includes a mixture of ethanol and methanol,
A
which could be separated using a second column, distillate product purity control is not aimed
M
in this study. Liquid levels are controlled by P-only controllers with a gain of 2. PI controllers
ED
including a single 60-s first-order measurement lag are used for temperature control loops. Controller tuning constants are calculated using ATV test and Tyreus–Luyben tuning rule
PT
embedded in the commercial software Aspen Dynamics. Temperature transmitter spans are selected 50 K, and all control valves are designed to be half open at steady state. Temperature
CC E
control trays are determined by performing open-loop sensitivity analysis. Then, disturbances such as changes in the production rate handle and the feed composition are applied to demonstrate their robustness and performance.
A
4.1 Control of RDC-SRS Schematic of the two-point temperature control structure proposed for RDC-SRS is illustrated in Figure 12. In this control structure, feed flowrate of ethanol is flow controlled and serves as the production rate handle. Reflux drum level and column base level are controlled by manipulating the distillate and bottoms flow rates, respectively. A ratio control keeps the reflux
flowrate in ratio with distillate flow rate. The condenser pressure is controlled by manipulating the condenser heat removal. According to the results of sensitivity analysis, tray 26 and tray 35 are selected the most sensitive locations for temperature control. Thus, reboiler heat duty and feed flowrate of dimethyl carbonate are arranged to control T26 and T35, respectively. This
IP T
control structure is labeled CS1R,S. Figure 13A gives the responses of CS1R,S to ±20% changes in production rate handle, FEtOH.
SC R
The solid lines illustrate the case of +20% change and the dashed lines illustrate the case of -
20% change. Results show that these disturbances can be easily handled with some oscillation. Both tray temperatures which are directly controlled settle down to their set points in 5 hours
U
by properly arranging the manipulated variables. In addition, the purity of diethyl carbonate,
N
which is the main product of the process, settles down to a new steady-state with fairly small
A
transient and steady-state deviations. The magnitude of the final steady-state deviation is
M
smaller than 0.2% for both disturbances in positive and negative directions.
ED
The response of CS1R,S to a 10% change in feed composition (XEtOH) is given in Figure 13B. Tray temperatures return to their set points in less than 4 hours. It also takes about 4 hours for
PT
diethyl carbonate purity to settle down to new steady-state with a quite small deviation. The
CC E
size of the deviation is less than 0.04%.
4.2 Control of RDC-DRS Figure 14 gives the control structure for RDC-DRS. The feed flowrate of ethanol in the upper
A
reactive section (FETOH1) is flow controlled and serves as production rate handle. The distillate and bottoms flow rates are manipulated to control the reflux drum level and column base level, respectively. A ratio control is used to keep the reflux flowrate in ratio with distillate flow rate. Condenser heat removal is manipulated to control the condenser pressure. As the result of the sensitivity analysis, T27 and T35 are controlled by manipulating the reboiler heat duty and feed
flowrate of dimethyl carbonate, respectively. In addition, a fixed ratio controller is added to the structure to satisfy the feed balance between FETOH1 and FETOH2 . This control structure is labeled CS1R,D. Figure 15A gives the responses of CS1R,D to changes in production rate handle, FETOH1. The
IP T
tray temperatures recovery back to their set points in around 2 hours. Although stable closedloop responses are obtained for this control structure, the product purity of diethyl carbonate in
SC R
the bottoms stream deviates slightly from its setpoint. Absolute steady-state deviation of diethyl
carbonate purity is less than 0.2% in the face of positive and negative disturbances. Figure 15B illustrates the responses of the CS1R,D to a 10% change in feed composition. It takes 3 hours for
U
the system to settle down to steady-state. Tray temperatures return to their set points, while the
N
change in diethyl carbonate purity is less than 0.01%. Table 4 gives the controller pairings and
M
A
tuning parameters of both control structures studied for the real system.
5. Conclusions
ED
In this paper, controllability of two reactive distillation column configurations including two stage consecutive reversible reactions is compared. First configuration is a conventional
PT
reactive distillation column with a single reactive section, while the second one is a novel
CC E
column with double reactive sections. Their control performances are evaluated based on two representative examples, a generic system and a real system for diethyl carbonate production. Since the neat reactive distillation columns act like pure integrators with respect to the reactants,
A
satisfying the desired conversion of reactants in the reactive zone is important, so they do not leave the system as impurity. For this reason, some tray temperatures in the reactive section should be effectively controlled to satisfy the necessary conversion. For that to happen, the temperature profile should be proper to inferentially control the temperature, which means there should be a notable temperature difference between adjacent trays. In this study, the
temperature difference between the trays of reactive section for the ideal system is much smaller than that of the real system. Thus, the results of generic system for both RDC-SRS and RDCDRS indicate that control structures including direct composition control besides the inferential temperature controllers are necessary to hold the product pruties at their specifications. On the other hand, using a control structure including only inferential temperature controllers is found
IP T
enough to provide an effective control for both RDC-SRS and RDC-DRS of the real system studied.
SC R
The steady-state deviations of product purities are given in Table 5, and these results
demonstrate that control performances of RDC-DRS are just as good as RDC-SRS for both reaction system. Thus, it can be concluded that the RDC-DRS can be an alternative to the
A
CC E
PT
ED
M
N
A
regard to steady-state and control performances.
U
conventional RDC-SRS for the separation of two-stage consecutive reversible reactions with
References (1) Luyben, W. L.; Yu, C. C. Reactive Distillation Design and Control, Wiley: New Jersey, 2008. (2) Ung, S; Doherty, M.F. Synthesis of reactive distillation systems with multiple equilibrium
IP T
chemical reactions, Industrial & Engineering Chemistry Research, 1995, 34, 2555-2565. (3) Luo, H. P.; Xiao, W.D. A reactive distillation process for a cascade and azeotropic reaction
SC R
system: Carbonylation of ethanol with dimethyl carbonate, Chemical Engineering Science, 2001, 56, 403-410.
(4) Hung, S. B.; Lai, I.; Huang, H. P.; Lee, M. J.; Yu C.C. Reactive distillation for two-stage
N
Chemistry Research, 2008, 47, 3076-3087.
U
reaction systems: Adipic acide and glutaric acid esterifications, Industrial & Engineering
A
(5) Orjuela, A.; Kolah, A.; Lira, A. T.; Miller, D. J. Mixed succinic Acid/acetic acid
ED
Research, 2011, 50, 9209-9220.
M
esterification with ethanol by reactive distillation, Industrial & Engineering Chemistry
(6) Keller, T.; Holtbruegge, J.; Gorak, A. Transesterification of dimethyl carbonate with
PT
ethanol in a pilot-scale reactive distillation column, Chemical Engineering Journal, 2012, 180, 309-322.
CC E
(7) Yao, X.; Huang, K.; Chen, H.; Li, S. Employing top−bottom recycled reactive distillation to the separations of adipic acid and glutaric acid esterifications, Industrial & Engineering Chemistry Research, 2013, 52, 16870-16879.
A
(8) Yu, C.; Yao, X., Huang, K.; Zhang, L.; Wang, S.; Chen, H. A reactive distillation column with double reactive sections for the separations of two-stage consecutive reversible reactions, Chemical Engineering and Processing, 2014, 79, 56-68.
(9) Hung, S. B.; Chen, J. H.; Lin, Y. D.; Huang, H. P.; Lee, M. J.; Ward, J. D.; Yu, C. C. Control of plantwide reactive distillation processes: Hydrolysis, transesterification and two-stage esterification, Journal of the Taiwan Instute of Chemical Engineers, 2010, 41, 382-402. (10)
Wei, H. Y.; Rokhmah, A.; Handogo, R.; Chien, I. Design and control of reactive
distillation process for the production of diethyl carbonate via two consecutive
(11)
IP T
transesterification reactions, Journal of Process Control, 2011, 21, 1193-1207.
Bo, C.; Tang, J.; Cui, M.; Feng, K.; Qiao, X.; Gao, F. Control and profile setting of
Engineering Chemistry Research, 2013, 52, 17465-17474. (12)
SC R
reactive distillation column for benzene chloride consecutive reaction system, Industrial &
Kaymak D. B.; Unlu, H.; Ofkeli, T. Control of a reactive distillation column with double
U
reactive sections for two-stage consecutive reactions, Chemical Engineering and
A
CC E
PT
ED
M
A
N
Processing, 2017, 113, SI, 86-93.
Figure Caption
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 1 – Flowsheets of reactive columns for ideal system: (A) RDC-SRS, (B) RDC-DRS
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 2 – Flowsheets of reactive columns for real system: (A) RDC-SRS, (B) RDC-DRS
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 3 – Temperature profiles: (A) Ideal system, (B) Real system
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 4 – Control structure of RDC-SRS for ideal system: CS1I,S
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 5 – Results of CS1I,S: ±20% FA
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 6 – Control structure of RDC-SRS for ideal system: CS2I,S
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 7 – Results of CS2I,S: (A) ±20% FA, (B) zA,B = 0.05
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 8 – Control structure of RDC-DRS for ideal system: CS1I,D
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 9 – Results of CS1I,D: ±20% FA
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 10 – Control structure of RDC-DRS for ideal system: CS2I,D
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 11 – Results of CS2I,D: (A) ±20% FA, (B) zA,B = 0.05
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 12 – Control structure of RDC-SRS for real system: CS1R,S
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 13 – Results of CS1R,S: (A) ±20% FEtOH, (B) zEtOH,DMC = 0.10
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 14 – Control structure of RDC-DRS for real system: CS1R,D
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Figure 15 – Results of CS1R,D: (A) ±20% FEtOH1, (B) zEtOH1,DMC = 0.10
A
CC E
PT
ED
M
A
N
U
SC R
IP T
Table 1. Kinetic and physical properties for the ideal example Parameter Value Column pressure (bar) 10 First stage activation energy (kcal mol-1) forward 30 reverse 30 -1 Second stage activation energy (kcal mol ) forward 30 reverse 30 First stage specific reaction rate at 366 K (kmol s-1 kmol-1) forward 0.008 reverse 0.004 Second stage specific reaction rate at 366 K (kmol s-1 kmol-1) forward 0.0008 reverse 0.0016 First stage chemical equilibrium constant at 366 K 2 Second stage chemical equilibrium constant at 366 K 0.5 -1 Heat of reaction (kcal mol ) 0 Heat of vaporization (kcal mol-1) 6.944 a Vapor pressure constants (AVP,j - BVP,j) A 10.96 – 3862 B 12.35 – 3862 C 11.65 – 3862 D 13.04 – 3862 E 10.27 – 3862 a ln P s A j VP , j BVP , j T with temperature in K and vapor pressure in bar
Value 1.013 9.553 9.553
0.0692 0.0346 0.0134 0.0298 2 0.45 0
IP T
8.904 8.904
A
CC E
PT
ED
M
A
N
U
SC R
Table 2. Kinetic and physical properties for the real example Parameter Column pressure (bar) First stage activation energy (kcal mol-1) forward reverse Second stage activation energy (kcal mol-1) forward reverse First stage specific reaction rate at 366 K (kmol s-1 kmol-1) forward reverse Second stage specific reaction rate at 366 K (kmol s-1 kmol-1) forward reverse First stage chemical equilibrium constant at 366 K Second stage chemical equilibrium constant at 366 K Heat of reaction (kcal mol-1)
A
CC E
PT
ED
M
A
N
U
SC R
τI (min) 62.33 17.60 62.33 32.27 9.99 11.64 14.39 9.99 11.64 28.88
IP T
Table 3. Controller tuning parameters for ideal system Control structure CV MV KC T17 FB 0.50 CS1I,S T20 R 0.56 T17 FB 0.50 CS2I,S xD,D R 4.38 T6 FB2 0.84 CS1I,D T13 FB1 6.57 T21 R 0.70 T6 FB2 0.84 CS2I,D T13 FB1 6.57 xD,D R 5.52
Table 4. Controller tuning parameters for real system Control structure CV MV KC T26 QR 0.41 CS1R,S T35 FDMC 6.70 T27 QR 0.42 CS1R,D T35 FDMC 13.31
A
CC E
PT
ED
M
A
N
U
SC R
IP T
τI (min) 10.56 11.62 12.27 11.75
Table 5- Steady-state deviations Ideal System CS2I,S
CS2I,D |∆XB,E| / |∆XD,D|
+20% in PRH
0.00000 / 0.00000
0.00000 / 0.00000
-20% in PRH
0.00000 / 0.00000
0.00003 / 0.00001
5% impurity in PRH
0.00003 / 0.00001
0.00000 / 0.00000
CS1R,S
CS1R,D
|∆XDEC|
|∆XDEC|
+20% in PRH
0.00155
0.00192
-20% in PRH
0.00106
0.00123
10% impurity in PRH
0.00034
0.00001
A
CC E
PT
ED
M
A
N
SC R
U
Real System
IP T
|∆XB,E| / |∆XD,D|