Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Chemical Engineering Science 61 (2006) 5377 5392 www.elsevier.

com/locate/ces

Towards further internal heat integration in design of reactive distillation columnsPart II. The process dynamics and operation
Kejin Huang a , Masaru Nakaiwa a, , Atsushi Tsutsumi b
a Research Institute for Innovation in Sustainable Chemistry, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba 305-8565, Japan b Department of Chemical System Engineering, The University of Tokyo, Tokyo 113-8565, Japan

Received 20 September 2004; received in revised form 30 November 2005; accepted 10 March 2006 Available online 17 March 2006

Abstract In the rst paper of this series, it has been demonstrated that the capital investment and operating cost can frequently be reduced substantially through seeking further internal heat integration between the reaction operation and separation operation for a reactive distillation column involving reactions with highly thermal effect. In this paper, the dynamics and operation of the resultant reactive distillation system is to be examined, with special emphasis focused on the dynamic effect of the supplementary internal heat integration. It has been found that seeking further internal heat integration can sometimes improve process dynamics and lessen difculties in process operation. This outcome stems from the rened relationship between the reaction operation and separation operation involved and is of great signicance in tightening process design for a reactive distillation column containing reactions with highly thermal effect. It should, however, be pointed out that seeking further internal heat integration might also conne severely the exibility of the resultant reactive distillation column due to the reduction of mass transfer driving forces. When encountering a sharp increase in the product specication that is more relevant to the supplementary internal heat integration, the process might show deteriorated dynamic performance and can even converge to an undesirable steady state where the economical advantages of the supplementary internal heat integration are lost totally. Therefore, some effective measures to increase the redundancy of the resultant process design have to be taken to deal with the side-effect during process development. 2006 Elsevier Ltd. All rights reserved.
Keywords: Distillation; Reaction engineering; Heat integration; System engineering; Process control; Dynamic simulation

1. Introduction For a reactive distillation column containing reactions with highly thermal effect, it has already been demonstrated that a substantial reduction of operating cost and capital investment can frequently be achieved by seeking further internal heat integration between the reaction operation and separation operation involved (Huang et al., 2005). Although the improvement in process design looks very attractive, the inuences to process dynamics and operation remain a matter of special concern. As the combination of the reaction operation and separation operation already causes a reactive distillation column more difcult to control than a conventional distillation column, it will

Corresponding author. Tel.: +81 029 861 4696; fax: +81 029 861 4660.

E-mail address: nakaiwa-m@aist.go.jp (M. Nakaiwa). 0009-2509/$ - see front matter doi:10.1016/j.ces.2006.03.015 2006 Elsevier Ltd. All rights reserved.

certainly perplex process designers if very complicated process dynamics and severe operation difculties can arise from this process innovation. It seems therefore imperative to investigate the detailed impact from the reinforcement of internal heat integration between the reaction operation and separation operation on process dynamics and operation. Thanks to the tremendous efforts exerted on reactive distillation systems worldwide so far, a deep insight has already been acquired into the process dynamics and operation. A great number of papers addressed the descriptions of reactive distillation operation by various approaches, which could generally be categorized into equilibrium and non-equilibrium cell models (Roat et al., 1986; Cuille and Reklaitis, 1986; Ruiz et al., 1995; Alejski and Duprat, 1996; Kreul et al., 1998; Baur et al., 2001; Peng et al., 2003). The possible existence of multiple steady states sparked a considerable interest in the mechanism behind this intricate phenomenon (Jacobs and Krishna,

5378

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

1993; Ciric and Miao, 1994; Hauan et al., 1995; Sneesby et al., 1998; Eldarsi and Douglas, 1998; Mohl et al., 1999; Guttinger and Morari, 1999a,b). The uncertainties introduced by the input and/or output multiplicities provoked serious concern on the smooth and reliable process operation and thus stimulated the development of robust control systems to deal with such complexities (Kumar and Daoutidis, 1999; Rosendo et al., 2000; Vora and Daoutidis, 2001). For the tight design of control systems, the complex process dynamics and strong nonlinearity had to be well compensated and this generally necessitated delicate modeling techniques and complicated control algorithms (Bisowarno et al., 2004). Recently, nonlinear model-based control algorithms were also employed to the operation of some reactive distillation columns (Balasubramhanya and Doyle III, 2000; Tian et al., 2003; Engell and Fernholz, 2003). Generally speaking, for the effective operation of a reactive distillation column, maintaining tightly product qualities and keeping a strict stoichiometric balance between reactants are the two essential objectives in the control system design. Although the product qualities can be controlled, in most cases, in a way exactly similar to the control of a conventional distillation column (e.g., by temperature or direct composition control), the proper control of the reactant stoichiometry appears to be quite troublesome. Ratio control system appears to be an appropriate candidate, however, the poor accuracy of ow measurements prohibits it from being an effective one (Chiang et al., 2002). Sneesby et al. (1999) proposed a two-point control scheme for an ethyl tert-butyl ether (ETBE) reactive distillation column in which both the product purity and the conversion rate could be controlled. An inferential model had, however, to be developed and employed to detect the degree of conversion within the reactive section. Luyben and his coworkers (Luyben, 2000; Al-Arfaj and Luyben, 2000) proposed and evaluated six alternative control structures for a hypothetical reactive distillation column. The use of a concentration analyzer in the reactive section was advocated to keep the stoichiometric balance between reactants. They further proved the necessity and the effectiveness of the internal composition control loop in the operation of an ETBE and a tert-amyl methyl ether (TAME) reactive distillation columns (Al-Arfaj and Luyben, 2002, 2004). Wang et al. (2003a,b) proposed a feed ratio plus internal concentration control scheme and found effective for two reactive distillation columns producing methyl tertiary butyl ether (MTBE) and n-butyl acetate, respectively. Except for the feedforward compensation effect involved, their control conguration worked actually in the same principle as that of Al-Arfaj and Luyben. Huang et al. (2004) devised a temperature plus feed ratio cascade control scheme for two heterogeneous reactive distillation columns synthesizing n-butyl propionate and butyl acetate, respectively. In order to deal with the variations in the production rate, they adjusted the setpoint of the temperature controller according to a simple regression model developed. With regard to process design and its inuences to process dynamics and operation, only a few papers have appeared on reactive distillation systems so far. Heath et al. (2000) discussed the selection of control structures in terms of an economical objective function for an ethylene glycol reactive distillation column.

Georgiadis et al. (2002) addressed the interaction between process design and control for a reactive distillation column synthesizing ethyl acetate and advocated the simultaneous consideration of process design and process operation in the early stage of process development. A common feature of these studies is the employment of a mixed-integer nonlinear programming (MINLP) approach. Although it is a very powerful synthesis method, it does not offer a clear picture that can reect the intricate relationship between process design and process dynamics and operation. For design of an intensied chemical process like a reactive distillation column, it is very helpful for a designer to know clearly some favorable design options and their detailed effect on process dynamics and operation in order to make an economical and yet reliable decision. Therefore, how to effectively tighten process design and meanwhile leave as small a negative effect as possible and even a positive effect to process dynamics and operation still remains to be an important issue to solve in the near future. The purpose of the present work is to evaluate the dynamic effect of the supplementary internal heat integration between the reaction operation and separation operation for a reactive distillation column involving reactions with highly thermal effect. After a brief introduction of the principle of internal heat integration, the process designs for two reactive distillation columns containing, respectively, a highly exothermic reaction and a highly endothermic one are described. Studies of process dynamics and operation are then performed on these two processes. Comparison of open-loop and closed-loop dynamic behaviors is conducted in depth between the process designs with and without further internal heat integration between the reaction operation and separation operation. Transitions to some undesirable steady states have been encountered in the servo responses and found to hold a close relationship with the supplementary internal heat integration. The implications of the reinforcement of internal heat integration are discussed and some concluding remarks are highlighted in the last section of the paper. 2. Process design by seeking further internal heat integration For a reactive distillation column containing reactions with highly thermal effect, the appropriate superimposition of reactive section onto rectifying/stripping section can serve to strengthen internal heat integration between the reaction operation and separation operation involved. There usually exist two design options in the process development. One is to extend reactive section to either rectifying section (in case of endothermic reactions) or stripping section (in case of exothermic reactions) in terms of the detailed reaction system at hand. The other is to choose deliberately the feed location of the relevant reactant in the reactive section. Starting from a basic process design with three distinct sections (i.e., rectifying, reactive and stripping sections), one can easily evolve it into a more efcient one with the combinatorial utilization of these two methods. More detailed information can be found in the rst paper of this series.

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392 Table 1 Physical properties and nominal operating condition of hypothetical reactive distillation columns Items Systems Example I Pressure (Bar) Overhead product composition (C, mole fraction) Bottom product composition (D, mole fraction) Holdup (kmol) Stage Condenser Reboiler Forward Backward Specic reaction rate at 366 K (kmol s1 kmol1 ) Relative volatility A:B:C:D Heat of reaction (kJ kmol1 ) Latent heat of vaporization (kJ kmol1 ) Vapor pressure constants A (Avp /Bvp ) B (Avp /Bvp ) C(Avp /Bvp ) D(Avp /Bvp ) Forward Backward 9 0.95 0.95 1 20 20 125 520 167 360 0.008 0.004 4:2:8:1 41 840 29053.7 12.3463/3862 11.6531/3862 13.0394/3862 10.96/3862 Example II 9 0.95 0.95
FA = 0.0126 kmol/s
14 2

5379

FB = 0.0126 kmol/s

RR = 0.033 kmol/s

d = 0.0126 kmol/s xC, d = 0.95

1 20 20 167 360 125 520 0.008 0.004 4:2:8:1 +41 840 29053.7 12.3463/3862 11.6531/3862 13.0394/3862 10.96/3862
FB = 0.0126 kmol/s

Q = 822.86 kW
21

Activation energy (kJ kmol1 )

(a)

b = 0.0126 kmol/s xD, bot = 0.95

RR = 0.0286 kmol/s d = 0.0126 kmol/s xC, d = 0.95

FA = 0.0126 kmol/s

13 14 17

Q = 695.942 kW

21

The two hypothetical reactive distillation systems, i.e., examples I and II, studied in the rst paper of this series, are employed here for examining the dynamic effect of the supplementary internal heat integration between the reaction operation and separation operation. In the reactive section of these reactive distillation columns, reactants A and B undergo a reversible liquid-phase reaction to form products C and D A + B C + D. (1)

(b)

b = 0.0126 kmol/s xD, bot = 0.95

Fig. 1. Process designs with and without further internal heat integration for example I: (a) 7/6/7; (b) 7/6(1)/7(3).

yi,j = xi,j Pis /Pj . The vapor saturation pressure is calculated as Ln Pis = Avp,i Bvp,i /Tj .

(3.2)

(4)

The volatilities are such that the products C and D are the lightest and heaviest, respectively, in the system. The net reaction rate for component i on stage j in the reactive section is given by ri,j = vi Hj (Kf,j xA,j xB,j Kb,j xC,j xD,j ), (2.1)

where Kf,j and Kb,j are the forward and backward specic reaction rates and given by Kf,j = Kb,j =
fe be Ef /(RT j )

(2.2) (2.3)

Eb /(RT j )

Ideal vapor and liquid phase behavior is assumed for the reaction system and the vaporliquid equilibrium relationship can be expressed as
s s s s Pj = xA,j PA + xB,j PB + xC,j PC + xD,j PD ,

(3.1)

Example I differs mainly from example II in the reaction kinetics. The former contains a highly exothermic reaction and the latter a highly endothermic one, both with an equal reaction heat. The detailed physical properties and nominal operating condition are shown in Table 1. Figs. 1 and 2 illustrate the basic process designs, 7/6/7s, and those with further internal heat integration between the reaction operation and separation operation, 7/6(1)/7(3) and 7(3)/6(2)/7, for examples I and II, respectively. Here, a unied notation, Nr (n1)/Nrea (n2)/Ns (n3), has been used to represent different process designs with and without further internal heat integration. Nr , Nrea , and Ns signify, respectively, the number of stages in the rectifying, reactive and stripping sections of the basic process design. The numbers in the parentheses, n1 and n3, stand for the superimposition of additional reactive stages onto the rectifying/stripping section, respectively, and n2, denotes the movement of the feed location

5380

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

1.01 1 xC, d (m.f.)


2

0.99 0.98 0.97 0.96 0.95

-10%

FB = 0.0126 kmol/s

RR = 0.0142 kmol/s

d = 0.0126 kmol/s xC, d = 0.95

+10% 0 2 4 Time [h] 6 8 10

FA = 0.0126 kmol/s

0.94
14

Q = 1278.026 kW
21

1.2
b = 0.0126 kmol/s xD, bot = 0.95

1 xD, bot (m.f.) 0.8 0.6 0.4 0.2

+10%

(a)

-10%

0
6 9

0
RR = 0.0095 kmol/s d = 0.0126 kmol/s xC, d = 0.95

4 Time [h]

10

FB = 0.0126 kmol/s

11

Fig. 3. Open-loop transient responses of the reactive distillation columns with and without further internal heat integration when they are subject to a 10% step change in the feed ow rate of reactant B, respectively (example I).

FA = 0.0126 kmol/s

14

Q = 1142.787 kW
21

(b)

b = 0.0126 kmol/s xD, bot = 0.95

Fig. 2. Process designs with and without further internal heat integration for example II: (a) 7/6/7; (b) 7(3)/6(2)/7.

of the relevant reactant within the reactive section. It is readily seen that seeking further internal heat integration has lead to a simultaneous reduction of capital investment and energy consumption. In the remainder of the paper, the dynamics and operation are to be examined for the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation. It is stipulated here that the solid lines represent the dynamic responses of the basic process designs, 7/6/7s, and the dashed lines the dynamic responses of the process designs, 7/6(1)/7(3) and 7(3)/6(2)/7. 3. Example I: a reactive distillation system involving a highly exothermic reaction 3.1. Open-loop transient responses to step changes in the feed ow rate In Fig. 3, the open-loop transient responses of the reactive distillation columns with and without further internal heat in-

tegration between the reaction operation and separation operation are depicted, when they are subject to a 10% step change in the feed ow rate of reactant B, respectively. After disturbed by a 10% step change in the feed ow rate of reactant B, the basic process design, 7/6/7, operates around the nominal steady state for only about an hour and then drifts away. Notice that the new steady state reached is far different from the nominal one with a steady state composition nearly pure for component C in the top product and close to zero for component D in the bottom product, implying, therefore, an abnormal situation with an extremely low conversion rate in the reactive section. Because of the slow composition dynamics, the sudden reduction in the feed ow rate of reactant B introduces no sharp changes in the composition prole within the reactive section and this is why the reactive distillation column can exhibit relatively small deviations from the nominal steady state for about an hour. The lack of reactant B leads to decrement in the conversion rate and affects gradually the internal vapor and liquid ow rates, thereby forcing the process to apart from the nominal operating condition. The variation of internal vapor and liquid ow rates accelerates further the runaway speed, as is clearly indicated by the turning point around the time of 3.2 h. The top product ow rate nally approaches to zero and makes the inventory control system of the reux-drum cease to function properly. For the process design with further internal heat integration between the reaction operation and separation operation, 7/6(1)/7(3), it is readily seen that the sensitivity to the feed ow rate disturbance has been suppressed, substantially. In addition to a much slower escaping speed, the

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

5381

0.98 0.975 0.97 xC, d (m.f.) 0.965 0.96 0.955 0.95 0.945 0.94 0 2 4 Time [h] 0.98 0.96 xD, bot (m.f.) 0.94 0.92 0.9 0.88 0.86 0.84 0 2 4 Time [h]
Fig. 4. Open-loop transient responses of the reactive distillation columns with and without further internal heat integration when they are subject to a 0.0001 kmol/s step change in the reux ow rate, respectively (example I).

0.98 0.97 xC, d (m.f.)


+0.0001 kmol/s

-0.0001 k mol/s

0.96 0.95 0.94 0 2 4 Time [h] 6 8 10 +0.0001 k mol/s

-0.0001 kmol/s

10

-0.0001 kmol/s

0.98 0.96 0.94 xD, bot (m.f.) 0.92 0.9 0.88 0.86 -0.0001 k mol/s 0.84 0.82 0 2 4 Time [h]
Fig. 5. Open-loop transient responses of the reactive distillation columns with and without further internal heat integration when they are subject to a 0.0001 kmol/s step change in the boilup ow rate, respectively (example I).

+0.0001 k mol/s

+0.0001 kmol/s 6 8 10

10

steady state composition of component D remains around 0.26 in the bottom product and is actually well above the corresponding value of the basic process design, 7/6/7, indicating a relatively higher conversion rate in the reactive section. There is no doubt that the improvement in process dynamics stems from the reinforcement of internal heat integration between the reaction operation and separation operation. It will certainly be favorable to process operation. When a +10% step change in the feed ow rate of reactant B has been encountered, fairly small differences have been observed between the process designs with and without further internal heat integration between the reaction operation and separation operation. Similar observations have also been made on the open-loop transient responses to the step changes in the feed ow rate of reactant A. 3.2. Open-loop transient responses to step changes in the reux ow rate The open-loop transient responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation are plotted in Fig. 4, after they are subject to a 0.0001 kmol/s step change in the reux ow rate, respectively. For the negative perturbation in the reux ow rate, there appears almost no sharp difference between the process designs, 7/6/7 and 7/6(1)/7(3), whereas for the positive perturbation in the reux ow rate, the former displays a much larger deviation than the latter in the top and the bottom products, thereby implying a signicant

departure from the nominal steady state. Although the reux ow rate has increased by only a small magnitude, it triggers a considerable change of the conversion rate in the reactive section, which is actually responsible for the runaway from the nominal operating condition. As the reux ow rate is the manipulative variable for the control of the top product, the great degree of asymmetry between the positive and the negative responses (i.e., process nonlinearity) may present severely detrimental inuences to the control system performance. With the introduction of further internal heat integration between the reaction operation and separation operation in the process design, 7/6(1)/7(3), the degree of the process asymmetry has been alleviated considerably. As can be seen in Fig. 4, the process is now able to operate in the vicinity of the nominal steady state and exhibits almost no noticeable differences between the positive and the negative responses, therefore signifying a sharp improvement in process dynamics. 3.3. Open-loop transient responses to step changes in the reboiler duty Fig. 5 shows the open-loop transient responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation, when they are subject to a 0.0001 kmol/s step change in the boilup ow rate, respectively. No sharp difference has been found between the process designs, 7/6/7 and 7/6(1)/7(3), in the presence of the positive perturbation in the boilup ow

5382

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

rate. However, a large degree of difference has been observed in the presence of the negative perturbation in the boilup ow rate. The basic process design, 7/6/7, exhibits again a very large shift from the nominal operating condition in a way extremely similar to the situation that a positive perturbation in the reux ow rate is confronted. With the consideration of the supplementary internal heat integration between the reaction operation and separation operation, the process design, 7/6(1)/7(3), is now able to work around the nominal steady state and the degree of the asymmetrical nonlinearity has been relieved substantially in comparison with that of 7/6/7. Since the boilup ow rate is the manipulative variable for the control of the bottom product, it is not difcult to understand that seeking further internal heat integration will be benecial to process operation. 3.4. Controllability analysis Here, controllability analysis is conducted in the frequency domain in terms of two predictive indexes: Morari resiliency index (MRI) and condition number (CN). The MRI is the minimum singular value of the open-loop transfer function, which corresponds actually to a specic input and output direction. The control system that presents a large MRI over the frequency range of interest is preferred. The CN is the ratio of the maximum singular value to the minimum singular value, which can work as an effective measure for system sensitivity under uncertainties in process parameters and modeling errors. The control system with a small CN over the frequency range
25 20 MRI 15 10 5 0 0.01 0.1 1 [rad/s]
900

CC

FC FB
9

LC

d, xC, d CC FA
14

21

LC CC b, xD, bot

Fig. 7. Control scheme for the reactive distillation column.

10

100

1000

of interest is preferred. These two controllability indexes were frequently applied to the evaluations of control system performance, especially in the development of thermodynamically efcient chemical processes through heat and/or mass integration (Shimizu et al., 1985; Skogestad et al., 1990; Huang et al., 2000; Jimenez et al., 2001; Serra et al., 2003). In Fig. 6, the dynamic effect of seeking further internal heat integration between the reaction operation and separation operation is demonstrated in terms of MRI and CN in the frequency domain. As can be seen, after the reinforcement of internal heat integration between the reaction operation and separation operation, the MRI has increased by a factor of 20 in the low frequency range. Even in the high frequency range, it has increased approximately by a factor of 3, although this tendency may not be identied clearly from the illustration. As far as the CN is concerned, a substantial reduction has resulted from the intensication of internal heat integration between the reaction operation and separation operation. It has contracted to about 10 in the process design, 7/6(1)/7(3), from the corresponding values of the basic process design, 7/6/7, i.e., around 800 in the low frequency range and 140 in the high frequency range. These outcomes predict obviously that seeking further internal heat integration is likely to present positive inuences to process dynamics and thus eases process operation for the reactive distillation system studied. 3.5. Closed-loop simulation

600 CN

300

0 0.01

0.1

10

100

1000

[rad/s]
Fig. 6. Positive effect of the supplementary internal heat integration on process controllability (example I).

Closed-loop simulation is conducted to further ascertain the dynamic effect of the supplementary internal heat integration between the reaction operation and separation operation. Al-Arfaj and Luyben (2000) once devised six control structures for the reactive distillation column, and we select the control scheme termed as CS1 by them in the following study. The control structure is reproduced in Fig. 7, where the purities of both the top and the bottom products are measured and controlled. In the top product, the composition of component C is controlled by manipulating the reux ow rate. In the bottom product, the composition of component D is controlled by

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

5383

0.97 0.96 xC, d (m.f.) 0.95 0.94 0.93

0.04 0.036 0.032 0.028 0.024

3 Time [h]

RR [kmol/s]

3 Time [h]

0.956 0.952 0.948 0.944

0.032 0.03 Vnt [kmol/s] 0 1 2 3 Time [h] 4 5 6

xD, bot (m.f.)

0.028 0.026 0.024 0.022

0.94

0.02 0 1 2 3 Time [h] 4 5 6

Fig. 8. Servo responses of the reactive distillation columns with and without further internal heat integration when the top control loop is subject to a 0.01 step change in setpoint, respectively (example I). Grey curves: 0.95 0.94; black curves: 0.95 0.96.

0.975 -20% 0.9625 RR [kmol/s] xC, d (m.f.)

0.045 +20% 0.035

0.95 0.9375 +20% 0.925 0 1 2 3 Time [h] 0.965 0.96 -20% 4 5 6

0.025 -20% 0.015 0 1 2 3 Time [h] 0.04 0.035 4 5 6

Vnt [kmol/s]

xD, bot (m.f.)

0.955 0.95 0.945 0.94 0.935 0 1 2 3 Time [h] 4 5 6 +20%

+20% 0.03 0.025 0.02 -20% 0.015 0 1 2 3 Time [h] 4 5 6

Fig. 9. Regulatory responses of the reactive distillation columns with and without further internal heat integration when production rate is subject to a 20% step change, respectively (example I).

5384

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

0.965 0.96 0.955 0.95 0.945 0.94 0 1 2 3 4 Time [h] 5 6 RR [kmol/s] xC, d (m.f.)

0.04

0.03

0.02

3 4 Time [h]

0.957 0.955 0.953 0.951 0.949

0.03

xD, bot (m.f.)

Vnt [kmol/s] 0 1 2 3 4 Time [h] 5 6

0.026

0.022

0.018

3 4 Time [h]

Fig. 10. Regulatory responses of the reactive distillation columns with and without further internal heat integration when FB is changed from a pure component ow of reactant B (zA,9 = 0.0 and zB,9 = 1.0) into a mixture ow of reactants A and B (zA,9 = 0.2 and zB,9 = 0.8) (example I).

manipulating the heat duty of reboiler. The concentration of reactant A on stage 14 in the reactive section is measured and controlled by manipulating the feed ow rate of reactant A (i.e., the so-called reactant stoichiometry control loop). The feed ow rate of reactant B is the production rate handle and is ow controlled. Because the production rate is strictly controlled in this control scheme, a fair comparison is allowed between the process designs with and without further internal heat integration between the reaction operation and separation operation. The levels of the reux-drum and the bottom reboiler are controlled by the distillate and the bottom product ow rates, respectively. The dynamics of concentration measurements is assumed to be two rst-order lags of 30 s in series. The transmitter span of all composition measurements is taken to be 0.1 and all control valves are designed to be half open at the nominal steady state. Proportional-only controllers are used for all level control loops and proportional plus integral (PI) controllers are adopted for the top and the bottom composition control loops. Although it is reasonable to use a PI controller in the reactant stoichiometry control loop, it may intensify the interaction between different control loops and thus worsen overall system performance. Since both the top and the bottom products are strictly controlled, the existence of a certain degree of discrepancy in the composition on stage 14 will not present strong inuences to the overall system performance. Hence, it seems more appropriate to use a proportional-only controller here. All composition controllers are tuned in terms of an experimental design approach proposed by Finco et al. (1989). The ultimate gain

and the ultimate frequency are found by increasing the gain of a proportional-only controller until sustained oscillations occur. Then, the ZieglerNichols settings are calculated for each control loop. Finally, a detuning factor f is searched for that makes all composition control loops have appropriate damping coefcients Kc = Kzn /f , TI = Tzn f , (5) (6)

where Kc and TI represent the proportional gain and integral time, respectively, and KZN and TZN denotes the ZieglerNichols settings. In Fig. 8, the dynamic responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation are illustrated, when the top control loop is subject to a 0.01 step change in setpoint, respectively. It is readily to see that the process design with further internal heat integration, 7/6(1)/7(3), presents improved responses in comparison with the basic process design, 7/6/7, especially in the case when the setpoint is increased from 0.95 to 0.96. In addition to the smoother transitions between different specications on the top product, the interaction to the bottom control loop has also been relieved, conrming denitely the positive effect of the supplementary internal heat integration on process dynamics and operation. At the new steady states, the former can still maintain higher thermodynamic efciency than the latter, thereby evidencing

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

5385

the robustness of the supplementary internal heat integration to the variations in operating conditions. Fig. 9 shows the dynamic responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation, when the production rate is subject to a 20% step change, respectively, in terms of a corresponding variation in the feed ow rate of reactant B. Again, it can be seen that the process design, 7/6(1)/7(3), outperforms 7/6/7, with relatively smaller maximum deviations and shorter setting time in both the top and the bottom control loops. The advantage in thermodynamic efciency has been kept at the new steady states in spite of a considerable change in the production rate. In Fig. 10, the dynamic responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation are displayed, when FB is changed from a pure component ow of reactant B (zA,9 = 0.0 and zB,9 = 1.0) into a mixture ow of reactants A and B (zA,9 = 0.2 and zB,9 = 0.8). It should be reminded here of the fact that this represents one of the most severe situations to check the static and dynamic performance of the supplementary internal heat integration. It is noted that very strong interaction has been observed between the top and the bottom control loops in the basic process design, 7/6/7. With the reinforcement of internal heat integration between the reaction operation and separation operation in the process design, 7/6(1)/7(3), far improved dynamic responses have been observed. The interaction has now been attenuated dramatically between the top and the bottom control loops besides the reduced setting time and maximum deviations. Even though a considerable change has taken place in the operating condition, seeking further internal heat integration still remains effective in reducing energy consumption when compared with the basic process design, 7/6/7. 4. Example II: a reactive distillation system involving a highly endothermic reaction 4.1. Open-loop transient responses to step changes in the feed ow rate In Fig. 11, the open-loop transient responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation are displayed, when they are subject to a 10% step change in the feed ow rate of reactant A, respectively. From the responses of the bottom product, it is hard to identify any differences between the process designs, 7/6/7 and 7(3)/6(2)/7. From the responses of the top product, a very large deviation has been observed in the former than in the latter, in the presence of the negative perturbation in the feed ow rate of reactant A. The gradual decline of the conversion rate in the reactive section initiates a substantial change of internal vapor and liquid ow rates, thereby resulting in nally a considerable shift from the nominal operating condition in the process design, 7/6/7. With the introduction of further internal heat integration between the reactive section and rectifying section in the pro-

1.05 0.9 xC, d (m.f.) 0.75 -10% 0.6 0.45 0 2 4 Time [h] 1.05 1 xD, bot (m.f.) 0.95 0.9 0.85 0.8 6 8 10

+10%

-10%

+10%

4 Time [h]

10

Fig. 11. Open-loop transient responses of the reactive distillation columns with and without further internal heat integration when they are subject to a 10% step change in the feed ow rate of reactant A, respectively (example II).

cess design, 7(3)/6(2)/7, the decline of the conversion rate has been, to a certain extent, suppressed and the steady state deviation has been reduced in the top product. The low sensitivity to the feed ow rate disturbance is certainly attributed to the reinforcement of internal heat integration and can exercise a benecial effect on process operation. Fig. 12 illustrates the open-loop transient responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation, when they are subject to a 10% step change in the feed ow rate of reactant B, respectively. As can be seen, the responses obtained are extremely analogous to those that are excited by the step changes in the feed ow rate of reactant A. Process design, 7(3)/6(2)/7, still exhibits a smaller steady state deviation than 7/6/7 in the top product, conrming again that the improvement in process dynamics originates from the reinforcement of internal heat integration between the reaction operation and separation operation. 4.2. Open-loop transient responses to step changes in the reux ow rate and reboiler duty As far as the open-loop transient responses to step changes in the reux and boilup ow rates are concerned, no signicant differences have been found between the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation.

5386
1 0.9

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

20 15 MRI
-10%

+10% xC, d (m.f.) 0.8 0.7

10 5 0 0.01

0.6 0.5

4 Time [h]

10

0.1

10

100

1000

[rad/s]
15

1.02 0.99
xD, bot (m.f.)

-10%

0.96 0.93 0.9 +10% 0.87 0.84 0 2 4 Time [h] 6 8 10

10 CN 5 0 0.01

0.1

10

100

1000

[rad/s]
Fig. 13. Almost no effect of the supplementary internal heat integration on process controllability (example II).

Fig. 12. Open-loop transient responses of the reactive distillation columns with and without further internal heat integration when they are subject to a 10% step change in the feed ow rate of reactant B, respectively (example II).

4.4. Closed-loop simulation 4.3. Controllability analysis The dynamic effect of seeking further internal heat integration between the reaction operation and separation operation is again investigated in terms of MRI and CN in the frequency domain and the detailed results are illustrated in Fig. 13. At rst glance, seeking further internal heat integration looks to give a negative impact to control system performance. However, one may nd that it is not the case by a closer examination. Although the MRI has decreased from 14.1 to 10.9 and the CN has increased from 6.8 to 8.8 in the low frequency range after seeking further internal heat integration, the magnitudes of changes are actually very small, implying probably trivial or even no degradation at all in system controllability. In the high frequency range, almost no differences have been observed between the process designs with and without further internal heat integration, 7/6/7 and 7(3)/6(2)/7. Because process controllability is more closely related to the high frequency behavior, it is therefore more reasonable to believe that seeking further internal heat integration gives almost no inuences to the controllability of the resultant process. Recall the fact that seeking further internal heat integration favors the rejection of disturbances from the feed ow rate of reactants A and B, it is not difcult to gure out that the dynamic performance of the resultant process is likely to be improved as compared with the basic process design, 7/6/7. The control system conguration is the same as is shown in Fig. 7. The dynamic responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation are shown in Fig. 14, when the bottom control loop is subject to a 0.01 step change in setpoint, respectively. As can be seen, a rather smoother transition has been accomplished by the process design, 7(3)/6(2)/7 than by 7/6/7, between different specications on the bottom product, with not only smaller overshoots but also shorter setting time. Furthermore, the interaction to the top control loop appears to be alleviated, substantially. At the new steady states reached the former can maintain its higher thermodynamic efciency than the latter. Fig. 15 depicts the dynamic responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation, when the production rate is subject to a 10% step change, respectively. Process design, 7(3)/6(2)/7, exhibits again much improved dynamic responses than 7/6/7, with not only smaller deviations but also shorter setting time. The advantage in thermodynamic efciency brought about by the supplementary internal heat integration has been retained at the new steady states reached. In Fig. 16, the dynamic responses of the reactive distillation columns with and without further internal heat integration between the reaction operation and separation operation are displayed, when FA is changed from a pure component ow of

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

5387

0.956 0.952 xC, d (m.f.) 0.948 0.944 0.94 0.936 0 1 2 3 4 Time [h] 5 6 RR [kmol/s]

0.016 0.014 0.012 0.01 0.008 0 1 2 3 4 Time [h] 5 6

0.97 0.96 xD, bot (m.f.) 0.95 0.94 0.93

0.052

Vnt [kmol/s]
0 1 2 3 Time [h] 4 5 6

0.046

0.04

0.034
0 1 2 3 4 Time [h] 5 6

Fig. 14. Servo responses of the reactive distillation columns with and without further internal heat integration when the bottom control loop is subject to a 0.01 step change in setpoint, respectively (example II). Grey curves: 0.95 0.94; black curves: 0.95 0.96.

0.965 -10% RR [kmol/s] xC, d (m.f.) 0.955

0.017 +10% 0.014

-10%

0.945 +10% 0.935

0.011

+10% -10%

0.008 0 1 2 3 Time [h] 4 5 6 0 1 2 3 Time [h] 0.05 -10% Vnt [kmol/s] 0.045 0.04 -10% 0.035 +10% 0.03 0 1 2 3 Time [h] 4 5 6 0 1 2 3 Time [h] 4 5 6 +10% 4 5 6

0.96 0.956 xD, bot (m.f.) 0.952 0.948 0.944 0.94

Fig. 15. Regulatory responses of the reactive distillation columns with and without further internal heat integration when production rate is subject to a 10% step change, respectively (example II).

reactant A (zA,14 = 1.0 and zB,14 = 0.0) into a mixture ow of reactants A and B (zA,14 = 0.95 and zB,14 = 0.05). Again, far stronger interaction has been observed between the top and the

bottom control loops in the basic process design, 7/6/7, than in the process design, 7(3)/6(2)/7. Seeking further internal heat integration can still secure a substantial reduction in energy

5388

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

0.952 0.95 0.948 0.946 0.944 0 1 2 3 Time [h] 0.9513 4 5 6 RR [kmol/s] xC, d (m.f.)

0.016 0.014 0.012 0.01 0.008

3 Time [h]

0.048 0.045 0.042 0.039

xD, bot (m.f.)

0.9492

0.9471

0.945 0 1 2 3 4 Time [h] 5 6

Vnt [kmol/s]

0.036 0 1 2 3 4 Time [h] 5 6

Fig. 16. Regulatory responses of the reactive distillation columns with and without further internal heat integration when FA is changed from a pure component ow of reactant A (zA,14 = 1.0 and zB,14 = 0.0) into a mixture ow of reactants A and B (zA,14 = 0.95 and zB,14 = 0.05) (example II).

consumption despite a considerable shift from the nominal design condition. 5. Discussions Al-Arfaj and Luyben (2000) examined systematically the control system design for example I and indicated that augment of liquid holdups (to say more exactly, the amount of catalyst employed) in the reactive section could improve the dynamic performance of the reactive distillation column. In terms of the present study, we may further systematize their conclusion: even though without increasing the amount of catalyst, the dynamic performance could still be enhanced by distributing appropriately the reactive section within the process. Not only the open-loop responses but also the closed-loop ones justify the possibility. The extension of their conclusion signies the vital importance of process design to process dynamics and operation. More specically, the improvement in process dynamics and operation can sometimes result from the elaboration of process design with almost no additional capital cost for a reactive distillation system involving reactions with highly thermal effect. It is worthwhile to look into the fact that seeking further internal heat integration can present positive inuences to the dynamics and operation of a reactive distillation column involving reactions with highly thermal effect. There is no doubt that the combination of the reaction operation and separation operation in a single unit leads to a more challenging problem in process operation than a conventional distillation column. The essential difculties arise actually from how to arrange this

combination in process design. It is one of the most important design objectives that affect not only process economics but also process controllability in all kinds of process intensications. Simply distributing the reactive section between the rectifying and stripping sections (e.g., in a basic process design) cannot always be the best design option, especially for some reactions with highly thermal effect. On the contrary, pursuing an optimum combination of the reaction operation and separation operation can sometimes gain simultaneous improvement in both aspects, because a more rened relationship has been developed between the reaction operation and separation operation. Seeking further internal heat integration serves to enhance the synergism between the reaction operation and separation operation and it is the primary reason that process economics and process controllability have been improved simultaneously for the two reactive distillation columns studied in this work. It must be pointed out here that seeking further internal heat integration between the reaction operation and separation operation might also give severe restrictions to the exibility of the resultant reactive distillation column. Figs. 17 and 18 present a detailed comparison of the servo responses to different magnitudes of step perturbations in the setpoints of control loops for examples I and II. Here, the step perturbations are made on the end products that are more relevant to the supplementary internal heat integration, i.e., the bottom product for example I and the top product for example II. It is noted that for the basic process designs, 7/6/7s, there appear no problems at all in the transitions between different product specications regardless of the magnitudes of the perturbations. However, for the process designs with further internal heat integration, 7/6(1)/7(3)

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

5389

0.98 0.96 RR [kmol/s] 0 1 2 3 Time [h]


0.98

0.05 0.045

xC, d (m.f.)

0.04 0.035 0.03 0.025

0.94 0.92 0.9 4 5 6

3 Time [h]

0.045 0.04

0.97

0.96

Vnt [kmol/s]

xD, bot (m.f.)

0.035 0.03 0.025 0.02

0.95

0.94

0.015 0 1 2 3 Time [h] 4 5 6 0 1 2 3 Time [h] 4 5 6

Fig. 17. Servo responses of example I when the bottom control loop is subject to a +0.01 and a +0.02 step changes in setpoint, respectively. Grey curves: 0.95 0.96; black curves: 0.95 0.97.

0.964 0.96 0.956 0.952 0.948 0.944 0 1 2 3 Time [h] 0.952 0.95 xD, bot (m.f.) 0.948 0.946 0.944 Vnt [kmol/s] 4 5 6 RR [kmol/s] xC, d (m.f.)

0.016 0.014 0.012 0.01 0.008 0 1 2 3 Time [h] 0.0468 4 5 6

0.0438

0.0408

3 Time [h]

0.0378 0

3 Time [h]

Fig. 18. Servo responses of example II when the top control loop is subject to a +0.005 and a +0.01 step change in setpoint, respectively. Grey curves: 0.95 0.955; black curves: 0.95 0.96.

5390

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

Table 2 Multiple steady states and the performance of reactive distillation systems Systems Items Process congurations Example I: xC,d = 0.95, xD,bot = 0.97 Example II: xC,d = 0.96, xD,bot = 0.95 7/6/7 7/6(1)/7(3) 7/6(1)/7(3) 7/6/7 7(3)/6(2)/7 7(3)/6(2)/7 Reux ow rate (mol/s) 34.7287 31.037 41.6447 14.9466 No solution 28.5072 Reboiler duty (kW) 871.46 764.17 1167.88 1305.82 No solution 1699.88 Features of steady states Nominal Desirable Undesirable Nominal Desirable Undesirable

0.957 0.956 xC, d (m.f.) 0.955 0.954 0.953 0.01

0.012

0.014 RR [kmol/s]

0.016

0.018

Fig. 19. Relationship between the reux ow rate and the top product composition when the bottom product is controlled at 0.95 for 7(3)/6(2)/7 of example II.

and 7(3)/6(2)/7, their dynamic responses appear to be quite sensitive to the magnitudes of the perturbations. For the relative small perturbations, they really exhibit better dynamic responses than the basic process designs, 7/6/7s, whereas for the relative large perturbations, they display much degraded ones. As can be seen, both the top and the bottom control loops behave very sluggishly after initial strong actions and leave large discrepancies for an extremely long period of time. The PI controllers increase slowly the reux and boilup ow rates in order to eliminate the discrepancies and drive nally these processes to two undesirable steady states. Table 2 gives a comprehensive comparison between the steady states for examples I and II. It can be seen that the process designs with further internal heat integration lose their advantages in thermodynamic efciency over the basic process designs at those undesirable steady states. Although there does exist a desirable steady state for the process design, 7/6(1)/7(3), the process fails to work around it during dynamic operation. For the process design, 7(3)/6(2)/7, even no desirable steady state can be found and this is well demonstrated by the relationship between the reux ow rate and the top product composition shown in Fig. 19. It indicates clearly that the top product cannot be concentrated to 0.96 with a smaller reux rate than that of the basic process design, 7/6/7. The failure to reach those desirable steady states is apparently related to the reinforcement of internal heat integration between the reaction operation and separation operation and therefore some effective measures must be taken to deal with this side-effect during process design. Remember the fact

that seeking further internal heat integration reduces actually the driving force of mass transfer in the rectifying/stripping section where internal heat integration has been arranged (Melles et al., 2000), it certainly becomes more difcult to reach a high degree of separation in process designs, 7/6(1)/7(3) and 7(3)/6(2)/7 than in 7/6/7s. This is why the transitions to uneconomical steady states occur in the dynamic operation. A direct method to solve this problem is therefore to make appropriate compensation for the loss in the driving forces of mass transfer by adding some separating stages in the corresponding section. An alternative method is to reduce somehow the degree of internal heat integration between the reaction operation and separation operation. Although no requirement is needed on capital investment, the degradation in thermodynamic efciency is inevitable. Therefore, the rst method is generally preferred to the second one in process development. In Figs. 17 and 18, the effect of adding three separating stages in process designs, 7/6(1)/10(3) and 10(3)/6(2)/7, is also demonstrated (represented here by the dotted lines). It is readily seen that the deciencies associated with the supplementary internal heat integration have been remedied, completely. Not only have the transitions to those undesirable steady states been avoided, but also the superiorities over the basic process designs, 7/6/7s, have been maintained in both static and dynamic responses. 6. Conclusion In addition to a substantial reduction of capital investment and operating costs, seeking further internal heat integration between the reaction operation and separation operation can sometimes improve process dynamics and alleviate difculties in process operation for a reactive distillation column involving reactions with highly thermal effect. As long as process design has been carried out adequately, the supplementary internal heat integration between the reaction operation and separation operation has been found to be relatively insensitive to the changes in operating conditions from the viewpoint of process dynamics and operation and this is in good accordance with the conclusion obtained in the rst paper of this series. The reason can solely be attributed to the rened relationship between the reaction operation and separation operation resulting from the reinforcement of internal heat integration. These ndings emphasize the utmost importance of tightening process design by means of strengthening the synergistic effect

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392

5391

within a reactive distillation column. Although these conclusions have been derived from the two hypothetical reactive distillation systems studied in this work, they are considered to be of great signicance and can be effective guidelines for the design and operation of some reactive distillation systems involving reactions with highly thermal effect. However, it must be borne in mind that seeking further internal heat integration between the reaction operation and separation operation might also conne severely the exibility of the resultant reactive distillation column due to the reduction of mass transfer driving forces. When a large increase has been encountered in the product specication that is more relevant to the supplementary internal heat integration, the resultant process design might show much deteriorated dynamic performance and can even approach to an undesirable steady state where the economical advantages of the supplementary internal heat integration have been lost totally. Therefore, some effective measures, e.g., adjustment of the degree of internal heat integration and addition of a number of separating stages in the rectifying/stripping section, must be taken to ensure the resultant process design with enough redundancy and exibility during process development. Current work is now underway to develop a systematic method that permits feasible process design for a reactive distillation column involving reactions with highly thermal effect.

T TI V x y z

temperature, K integral time, s vapor ow rate, kmol s1 liquid composition vapor composition feed composition

Greek letters v pre-exponential factor stoichiometric coefcients of a reaction frequency, rad s1

Subscripts A b bot B C d D f i j nt r rea s ZN component index backward reaction bottom product component index component index distillate component index forward reaction component index stage index total number of stages rectifying section reactive section stripping section ZieglerNichols settings

Notation A Avp b B Bvp C CC CN d D E f F FC H K Kc L LC MRI n N P PI Q r R RR reactive component vapor pressure constant, Pa bottom withdrawal, kmol s1 reactive component vapor pressure constant, Pa K product component composition controller condition number distillate ow rate, kmol s1 product component activation energy of a reaction, kJ kmol1 detuning parameter feed ow rate of reactants, kmol s1 ow rate controller holdup, kmol specic reaction rate, kmol s1 kmol1 proportional gain liquid ow rate, kmol s1 level controller Morari resilience index number of stages number of stages pressure, kPa proportional plus integral controller heat duty, kW net reaction rate, kmol s1 ideal gas law constant, kJ kmol1 K1 reux ow rate, kmol s1

Superscripts s saturation

Acknowledgments A dynamic model of the hypothetical reactive distillation column involving a highly exothermic reaction is provided by Prof. W. L. Luyben and Prof. M. A. Al-Arfaj and hereby is acknowledged. The authors are grateful to the nancial support from the Japan Science and Technology (JST) Corporation under the frame of Core Research and Evolutional Science and Technology (CREST). References
Al-Arfaj, M.A., Luyben, W.L., 2000. Comparison of alternative control structures for an ideal two-product reactive distillation column. Industrial Engineering and Chemistry Research 39, 32983307. Al-Arfaj, M.A., Luyben, W.L., 2002. Control study of ethyl tert-butyl ether reactive distillation. Industrial Engineering and Chemistry Research 41, 37843796. Al-Arfaj, M.A., Luyben, W.L., 2004. Plantwide control for TAME production using reactive distillation. A.I.Ch.E. Journal 50, 14621473. Alejski, K., Duprat, F., 1996. Dynamic simulation of the multi-component reactive distillation. Chemical Engineering Science 51, 42374252. Balasubramhanya, L.S., Doyle III, F.J., 2000. Nonlinear model-based control of a batch reactive distillation column. Journal of Process Control 10, 209218. Baur, R., Taylor, R., Krishna, R., 2001. Dynamic behavior of reactive distillation columns described by a non-equilibrium stage model. Chemical Engineering Science 56, 20852102.

5392

K. Huang et al. / Chemical Engineering Science 61 (2006) 5377 5392 Luyben, W.L., 2000. Economic and dynamic impact of the use of excess reactant in reactive distillation systems. Industrial Engineering and Chemistry Research 39, 29352946. Melles, S., Grievink, J., Schrans, S.M., 2000. Optimization of the conceptual design of reactive distillation columns. Chemical Engineering Science 55, 20892097. Mohl, K.D., Kienle, A., Gilles, E.D., Rapmund, P., Sundmacher, K., Hoffmann, U., 1999. Steady state multiplicities in reactive distillation columns for the production of fuel ethers MTBE and TAME: theoretical analysis and experimental verication. Chemical Engineering Science 54, 10291043. Peng, J., Edgar, T.F., Eldridge, R.B., 2003. Dynamic rate-based and equilibrium models for a packed reactive distillation column. Chemical Engineering Science 58, 26712680. Roat, S.D., Downs, J.J., Vogel, E.F., Doss, J.E., 1986. The integration of rigorous dynamic modeling and control system synthesis for distillation columns: an industrial approach. In: Morari, M., McAvoy, T.J. (Eds.), Chemical Process ControlCPC III. Elsevier, New York. Rosendo, M.L., Eduardo, P.C., Jose, A.R., 2000. A robust PI control conguration for a high-purity ethylene glycol reactive distillation column. Chemical Engineering Science 55, 49254937. Ruiz, C.A., Basualdo, M.S., Scenna, N.J., 1995. Reactive distillation dynamic simulation. Transactions of the Institution of Chemical Engineers Part A 73, 363378. Serra, M., Espuna, A., Puigjaner, L., 2003. Controllability of different multicomponent distillation arrangements. Industrial Engineering and Chemistry Research 42, 17731782. Shimizu, K., Holt, B.R., Morari, M., Mah, R.S.H., 1985. Assessment of control structures for binary distillation columns with secondary reux and vaporization. Industrial and Engineering Chemistry Process Design and Development 24, 852858. Skogestad, S., Jacobsen, E.W., Morari, M., 1990. Inadequacy of steadystate analysis of feedback control: distillate-bottom control of distillation columns. Industrial Engineering and Chemistry Research 29, 23392346. Sneesby, M.G., Tade, M.O., Smith, T.N., 1998. Steady state transitions in the reactive distillation of MTBE. Computers and Chemical Engineering 22, 879892. Sneesby, M.G., Tade, M.O., Smith, T.N., 1999. Two-point control of reactive distillation column for composition and conversion. Journal of Process Control 9, 1931. Tian, Y.C., Zhou, F.T., Bisowarno, B.H., Tade, M.O., 2003. Pattern-based predictive control for ETBE reactive distillation. Journal of Process Control 13, 5767. Vora, N., Daoutidis, P., 2001. Dynamics and control of an ethyl acetate reactive distillation column. Industrial Engineering and Chemistry Research 40, 833849. Wang, S.J., Wong, D.S.H., Lee, E.K., 2003a. Effect of interaction multiplicity on control system design for a MTBE reactive distillation column. Journal of Process Control 13, 503515. Wang, S.J., Wong, D.S.H., Lee, E.K., 2003b. Control of a reactive distillation column in the kinetic regime for the synthesis of n-butyl acetate. Industrial Engineering and Chemistry Research 42, 51825194.

Bisowarno, B.H., Tian, Y.C., Tade, M.O., 2004. Adaptive control of an ETBE reactive distillation column. Journal of Chemical Engineering of Japan 37, 210216. Chiang, S.F., Kuo, C.L., Yu, C.C., Wong, D.S.H., 2002. Design alternatives for the amyl acetate process: coupled reactor/column and reactive distillation. Industrial Engineering and Chemistry Research 41, 32333246. Ciric, A.R., Miao, P., 1994. Steady state multiplicities in an ethylene glycol reactive distillation column. Industrial Engineering and Chemistry Research 33, 27382748. Cuille, P.E., Reklaitis, G.V., 1986. Dynamic simulation of multi-component batch rectication with chemical reactions. Computers and Chemical Engineering 10, 389398. Eldarsi, H.S., Douglas, P.L., 1998. Methyl tert-butyl ether catalytic distillation column: multiple steady states. Transactions of the Institution of Chemical Engineers Part A 76, 509516. Engell, S., Fernholz, G., 2003. Control of a reactive separation process. Chemical Engineering and Processing 42, 201210. Finco, M.V., Luyben, W.L., Polleck, R.E., 1989. Control of distillation columns with low relative volatilities. Industrial Engineering and Chemistry Research 28, 7583. Georgiadis, M.C., Schenk, M., Pistikopoulos, E.N., Gani, R., 2002. The interaction of design, control and operability in reactive distillation systems. Computers and Chemical Engineering 26, 735746. Guttinger, T.E., Morari, M., 1999a. Predicting multiple steady states in equilibrium reactive distillation. I. Analysis of nonhybrid systems. Industrial Engineering and Chemistry Research 38, 16331648. Guttinger, T.E., Morari, M., 1999b. Predicting multiple steady states in equilibrium reactive distillation. II. Analysis of hybrid systems. Industrial Engineering and Chemistry Research 38, 16491665. Hauan, S., Hertzberg, T., Lien, K.M., 1995. Why methyl tert-butyl ether production by reactive distillation may yield multiple solutions. Industrial Engineering and Chemistry Research 34, 987991. Heath, J.A., Kookos, I.K., Perkins, J.D., 2000. Process control structure selection based on economics. A.I.Ch.E. Journal 46, 19982016. Huang, K., Qian, J., Nakaiwa, M., Takamatsu, T., 2000. Assessment of control congurations for a general heat-integrated distillation column. Chinese Journal of Chemical Engineering 8, 339346. Huang, S.G., Kuo, C.L., Hung, S.H., Chen, Y.W., Yu, C.C., 2004. Temperature control of heterogeneous reactive distillation. A.I.Ch.E. Journal 50, 22032216. Huang, K., Iwakabe, K., Nakaiwa, M., Tsutsumi, A., 2005. Towards further internal heat integration in design of reactive distillation columnsPart I: the design principle. Chemical Engineering Science 60, 49014914. Jacobs, R., Krishna, R., 1993. Multiple solutions in reactive distillation for methyl tert-butyl ether synthesis. Industrial Engineering and Chemistry Research 32, 17061709. Jimenez, A., Hernandez, S., Montoy, F.A., Zavala-Garcia, M., 2001. Analysis of control properties of conventional and non-conventional distillation sequences. Industrial Engineering and Chemistry Research 40, 37573761. Kreul, L.U., Gorak, A., Dittrich, C., Barton, P.I., 1998. Dynamic catalytic distillation: advanced simulation and experimental validation. Computers and Chemical Engineering 22, S371S378. Kumar, A., Daoutidis, P., 1999. Modeling, analysis and control of ethylene glycol reactive distillation column. A.I.Ch.E. Journal 45, 5168.

You might also like