Widening the Applicability of Reactive Distillation Technology by

F. Citro and J. W. Lee* ... are based on the identification of the process dominant variables using learning ... Reactive distillation for producing m...
0 downloads 0 Views 179KB Size
Download PDF
Ind. Eng. Chem. Res. 2004, 43, 375-383

375

Widening the Applicability of Reactive Distillation Technology by Using Concurrent Design F. Citro and J. W. Lee* Department of Chemical Engineering, The City College of the City University of New York, New York, New York 10031

In this paper the concurrent design method based on partial control, which was developed by Shinnar and co-workers [(Ind. Eng. Chem. Res. 1995, 34, 1228; 1995, 34, 3014; 1996, 35, 2215; 1997, 36, 747; 2000, 39, 103) (AIChE J. 2000, 46, 2456)] is applied to reactive distillation systems. It is shown that this approach provides important insights on the design and control of such processes. One main advantage of the method is that the design and scale-up of complex processes are based on the identification of the process dominant variables using learning models such as literature, pilot plants, and simulation models. While we use the methyl acetate production system as our prime example, the results could be of general interest. Our results also explain what are the limitations in the current application of reactive distillation and what could be done to increase its applicability to a wider set of problems. 1. Introduction The subject of this paper is the study of reactive distillation to widen the applicability of the technology by overcoming its current limitations. Reactive distillation has already been in practice for over 20 years but it had only a few spectacular successes such as the production of methyl acetate. The number of actual installations is quite small despite a large number of patents and extensive academic and industrial research in the area. This paper tries to elucidate the reason for that. The main reason is that it is difficult to match the optimum conditions for distillation and reaction. In distillation, pressure is the only variable that can change separation temperature. Regrettably, many distillation processes are very inflexible in the conditions required as boiling points change with pressure and therefore the corresponding temperature has a narrow range. The conditions for reaction, on the other hand, are fixed by the properties of available catalysts. In this sense reactive distillation shows the general problem of process intensification: when carrying out two tasks in the same unit, it often involves compromises between reaction and separation conditions that negate the cost advantages gained by the intensification. Thus, for many interesting cases the optimum temperature for the catalyst is usually different from that required for the distillation. Also, low space velocities are difficult to achieve inside a distillation column and some standard ways of compensating for the deactivation of catalysts, such as raising the temperature or frequent catalyst regeneration, are not feasible inside the distillation column. Therefore, the only successful realizations are those to which the above limitations do not apply. On the other hand, reactive distillation has very significant potential advantages for cases involving azeotropes or complex separations. For example, it permits the design of countercurrent reactors with two * To whom correspondence should be addressed. Tel.: 212650-6688. Fax: 212-650-6660. E-mail: [email protected].

miscible reactants. This paper suggests and analyzes design modifications that allow achieving the full advantages of reactive distillation while keeping distillation and reaction in separate vessels. To develop our ideas and demonstrate them, we use the methyl acetate production system as an example. Reactive distillation for producing methyl acetate reduced investment and operative costs by almost an order of magnitude as it eliminated a whole series of separation processes. Since the methyl acetate process has a very active homogeneous catalyst, it can easily be realized in a single reactive column. The cost reduction is far larger than what could be expected with process intensification. The paper, therefore, analyzes the design and development of this process from a reaction engineering point of view. The reaction of methanol with acetic acid has an equilibrium constraint that cannot be overcome in the standard way of coupling the reactor with a distillation column because of the methyl acetate-methanol and methyl acetate-water azeotropes. The existence of these two azeotropes requires a very complex separation process. The countercurrent reactive distillation column eliminates the methyl acetate-methanol azeotrope by achieving complete conversion of methanol and breaks the methyl acetate-water one by extractive distillation. We will show that the same advantage can also be achieved by combining the distillation column with four external reactors. But this is more expensive than a single reactive column. However, it still retains the large economic advantages over the conventional process. We will use here the methyl acetate process because there are sufficient data available for reliable simulations. The results of this paper provide guidelines for those who want to investigate if reactive distillation is suitable for their processes or what potential advantages there could be. This could lead to successful efforts that could allow applications of this powerful technology to a much wider set of complex processes. To achieve this goal, we will first review the concurrent design method and derive the external-reactor column by generating the design alternatives of the methyl acetate production system. Then, we will identify

10.1021/ie030172f CCC: $27.50 © 2004 American Chemical Society Published on Web 08/28/2003

376

Ind. Eng. Chem. Res., Vol. 43, No. 2, 2004

the dominant variables of this system by applying the concurrent design. Finally, we will compare the process performance between the conventional reactive column and the external-reactor column in terms of the dominant variables. 2. Technical Background In recent years, Shinnar and his research group1-6 have proposed a concurrent design technique based on partial control for complex chemical processes. The main idea of the concurrent design is that even very complex systems can be safely built if the designs of both the process and its control structure are carried out simultaneously. Historically, this idea was introduced when the research focused its interest on the selection of control structures in plantwide control.7-14 More detailed references can be found in the review papers.15,16 In complex chemical systems, the number of process variables to be controlled is usually greater than the number of available manipulated variables. The central premise of partial control is that if we control a subset of process variables that greatly affect the performance of the process, we can maintain the other process variables within their desired ranges. The variables in this subset are called dominant variables.3,17 The primary goal of a plant is to produce products within desired specifications. If we do not meet the specifications, the plant fails. There are two types of specifications, product specifications (e.g., conversion, purity, and product properties) and process specifications (e.g., productivity, adaptability to different operating conditions, and change of feed). The specifications must be met without violating process constraints, which generally regard catalyst properties, process kinetics, materials of construction, gas effluents concentration, margin of action of manipulated variables, etc. The specifications are not given as a single point but as a range and their desired value may change due to market demands. Controlling the dominant variables enables us to achieve plant operations within the desired product and process specifications. The dominant variables are identified through a full understanding of the process. This identification can be done by using learning models such as bench scale or pilot plants and computer simulation models. Especially, sensitivity analysis is very useful for determining dominant variables in computer simulations. Once the dominant variables are identified, we determine the practical degrees of freedom, which are the number of dominant variables that can be independently controlled.5,6 Then, we assign available manipulated variables to the independent dominant variables and ensure that their pairing allows for sufficient control performance in terms of on-spec production and stabilization of the process. If this sufficiency is not achieved, we should modify the design to achieve it. A detailed procedure to implement the concurrent design can be seen in Shinnar et al.5 while specific mathematical and thermodynamic formulations can be found in Kothare et al.6 and Tyre´us.18,19 3. Partial Control and Concurrent Design of the Methyl Acetate Production System Under some circumstances, reaction and distillation can be integrated in a single operation unit. Such integration results in dramatic economic effects in terms

Figure 1. Conventional reactor with recycle from a distillation column. Table 1. Effect of Pressure on the Boiling Points T at 1 atm (°C) methyl acetate methanol water acetic acid sulfuric acid

Pure Components 57.1 64.5 100.0 117.9 274.8

methyl acetate-methanol methyl acetate-water

Azeotropes 53.9 56.1

T at 10 atm (°C) 143.6 137.3 180.5 213.1 344.6 130.2 139.5

of process miniaturization and simplification. For example, one reactor and nine separation columns were involved in a conventional process to produce methyl acetate. But in the early 1980s, Eastman Chemicals developed and built a single reactive distillation column that did the same tasks as those 10 operation units.20,21 It was reported that the operating and investment costs were reduced to one-fifth of the costs of the conventional process.22 Since this process came out, both academia and industry have been actively involved in applying reactive distillation technology to other systems. Detailed reviews on reactive distillation can be found in Buzad and Doherty and Buzad,23 Taylor and Krishna,24 and Malone and Doherty.25 3.1. Methyl Acetate Reaction System. Methyl acetate (MA) is produced via esterification of acetic acid (AC) with methanol (MT). The reaction takes place in the presence of sulfuric acid (SA) as a catalyst and forms water (W) as a byproduct: AC + MT S MA + W. Recovery of pure MA by distillation is very difficult due to the presence of two minimum boiling azeotropes. One azeotrope is MA-W (5 wt % of W at 1 atm) and the other is MA-MT (18 wt % of MT at 1 atm). Furthermore, a tangent pinch between AC and W makes the recovery of unconverted AC very expensive. Table 1 shows boiling points for pure components and azeotropes at two different pressures. AC is the highest boiler while the MA-MT azeotrope is the lowest boiler. 3.2. Generating Design Alternatives. A conventional scheme for producing MA would be a reactor followed by a nonreactive distillation column as shown in Figure 1. An excess of one of the two reactants is needed to achieve high reaction conversion. However, no matter what excess of AC or MT is used, pure MA cannot be separated at the top of the column due to the two minimum boiling azeotropes (MA-MT and MA-W).22 The only way to improve the product purity up to the values required by market demands is to circumvent the azeotropes. In this example, both the azeotropes can be overcome. Specifically, the MA-MT azeotrope is circumvented by reacting away MT, which results in a very high reaction conversion (over 99%). On the other

Ind. Eng. Chem. Res., Vol. 43, No. 2, 2004 377

Figure 2. Countercurrent scheme with separation in between. The reaction only occurs in the reactors (R1, R2, and R3). DC1, DC2, and DC3 are nonreactive distillation columns.

hand, the MA-W azeotrope can be broken by extractive distillation. Luckily enough, one of the reactants, AC, works as an agent for water extraction. A scheme that would be able to get rid of both azeotropes, while producing high-purity MA, requires the use of several reactors connected to each other in a countercurrent manner as shown in Figure 2. In common practice, countercurrent schemes are usually used when the systems are immiscible as in L-L extraction processes. Since the methyl acetate system is miscible, we have to put separation units between the reactors. AC is fed to an extractive distillation column where it breaks the methyl acetate-water azeotrope, and MT is provided at the opposite end. In this way, the final conversion would be higher than it is in the single reactor of Figure 1 because the reaction takes place with a large excess of one of the two reactants in each reactor. Pure methyl acetate is the top product of the extractive distillation column while acetic acid excess leaves the process with byproduct W and SA catalyst. The problem with this scheme is that it would be very expensive to recover excess AC from W. What can be observed in Figure 2 is that all the columns and reactors are fed with the same components but their concentrations change along the countercurrent train. This feature suggests to us that all the columns can be put together to form a single separation column as shown in Figure 3a. From this single column it is possible to withdraw side streams into external reactors and to return the reacted streams to the column in a manner similar to the countercurrent scheme in Figure 2. The two reactants are still fed countercurrently in Figure 3a while the SA catalyst is only provided to the first side reactor since it flows down with the liquid stream into the other reactors. A further improvement in the design is achieved when the reactors are incorporated inside the column and both tasks, distillation and reaction, are performed simultaneously. There are some advantages of this process intensification. The most interesting one is that we can build a more compact plant using a smaller number of units and consequently save installation and operating costs. Also, with this design it is possible to break the azeotropes both by complete conversion of one of the reactants and by extractive distillation. Finally, we achieve a reaction extent larger than equilibrium would permit. In fact, due to the evaporation occurring at each stage, MA (the lowest boiling component) is continuously removed from the liquid phase, where the reaction takes place, and more reactant can be converted.

Figure 3. Two schemes for methyl acetate reactive distillation. (a) Side reactors cascaded into a column. (b) Conventional reactive distillation column.

However, this process intensification is not feasible in general as it can be applied only to those processes that satisfy two conditions as the MA production system does. That is, reaction temperature has to match distillation temperature and there is to be a catalyst that is active at the matched temperature. In fact, separation requirements not only dictate the column temperature profile but also fix the internal liquid and vapor flow rates. With respect to this latter feature, a homogeneous catalyst allows us to freely adjust the space velocity (defined as volumetric feed flow rate/volume of catalyst) by simply changing the catalyst flow rate. Furthermore, the catalyst is continuously replaced because it leaves the column with the reaction products. In contrast, a heterogeneous catalyst would pose some problems. First of all, on-line regeneration and replacement are not possible and thus costly shutdowns and start-ups have to be done frequently. Since the internal flows are fixed by the separation requirements, and the catalyst holdup is fixed during the design stage, we lose a degree of freedom, the controllability of space velocity. To provide a sufficient residence time and guarantee that the catalyst remains active for an acceptably long time, complex internal designs and large catalyst hold-ups (tall column) are to be provided. These limitations strongly reduce the applicability of reactive distillation columns. It is shown later on in this paper how a design change can overcome all the above limitations and widen the use of this technique. In summary, we show how, starting from the most conventional reactive scheme and applying this design procedure, we end up with the countercurrent reactive distillation column used by Eastman Chemicals for the MA production in Figure 3b. 3.3. Specifications and Constraints. As already mentioned, product specifications are the most important specifications since off-spec products cannot be sold in the market. Here, the main product specification is

378

Ind. Eng. Chem. Res., Vol. 43, No. 2, 2004

MA purity at the top and it is required to be at least 99 mol % (or 99.5 wt %).21 Hereafter, the MA purity is expressed as mol % throughout the article. This value can change according to different market demands. On the other hand, an important process specification is the production rate (productivity) of MA or the distillate flow rate (D) to timely supply the required quantity of the product to the market. The set-point of the distillate flow rate is fixed by the feed flow rates of MT and AC or their ratio. The usage of SA causes a corrosion problem, specifically at the bottom where the temperature in the column is the highest. To minimize the corrosion problem, more expensive materials can be used and/or the bottom temperature can be maintained as low as possible. 3.4. Identification of Dominant Variables. Dominant variables are identified through sensitivity analyses in Aspen Plus simulations and by referring to the literature.20,21,26-29 The dominant variables causing multiple steady states were found by the Aspen Plus simulations (version 11.1, property set: WILSONNTH). For the detailed simulation results, refer to Citro and Lee.31 3.4.1. Column Pressure. Similarly to conventional distillation, pressure is dominant for a reactive distillation column since it affects the boiling points of the pure components and azeotropes, given in Table 1. At atmospheric pressure, MA has a lower boiling point than MT. However, at a pressure of 10 atm, the boiling point of MT is lower than that of MA. Consequently, pure MA cannot be produced at the top. Thus, the column pressure must be kept close to 1 atm. This means that the column pressure cannot be used as a manipulated variable in the partial control scheme. However, the pressure has to be controlled for inventory purposes. Here, we manipulate the flow rate of inert N2 to control the column pressure. Its dynamic response to the gas flow rate of N2 is quite fast. 3.4.2. Reflux Ratio. One striking feature of reactive distillation is that there is not only a minimum but also a maximum reflux ratio for feasibly achieving high reaction conversion and top product purity. From pilotplant experiments, Agreda and Partin20 reported that reaction conversion decreases at reflux ratios above 2.0. In the actual commercial running, the optimum operating range is 1.65 to 1.85 depending on catalyst activity, required production rate, and product purity.21 This optimal range of the reflux ratios was verified by visualization in composition space.26 One qualitative explanation for the existence of this optimal range is the following: If the reflux ratio decreases below the minimum, then AC can appear at the top product due to poor separation. If the reflux ratio is higher than the upper bound, AC comes out at the bottom due to the improved separation of unconverted AC. Thus, the reflux ratio is one of dominant variables since it affects reaction as well as separation performance. In general, an optimal range for the reflux ratio can be found in quaternary reactive systems with medium boiling products (saddle points) such as esterification systems.27,28 3.4.3. Temperature Profile. The column temperature profile plays an even more crucial role in reactive distillation than in conventional distillation. In fact, similarly to conventional distillation systems, the temperature is dominant for separation. But, as already explained, the separation also prescribes reaction tem-

perature. Hence, the column temperature profile is also dominant for reaction equilibrium and reaction rate. 3.4.4. Feed Flow Rates (Feed Ratios). The feed flow rates of MT and AC, like any other internal flow (i.e., reflux flow), are dominant for the process because they can change desired outputs, for example, column composition profile. However, we can only manipulate one feed flow rate in excess since the value for the limiting reactant is imposed on the column by the market to guarantee the desirable productivity (D). The set-points of feed flow rates are kept constant and are only adjusted when required to compensate for changes in market demands or in feed compositions. 3.4.5. Composition Profile. The column composition profile is also dominant since it can affect the desired purity of MA and the direction of reaction (forward or reverse). The composition profile is controlled by the feed ratio of MT and AC. 3.4.6. Catalyst Activity. Catalyst activity is dominant because it can change the reaction rate and therefore affect reaction conversion. Since the catalyst used for the methyl acetate production system is liquid SA, its activity hardly decays. To achieve a higher reaction conversion and corresponding higher MA purity (in other words, to increase the catalyst activity), the flow rate of SA can be increased.21 However, high SA concentration causes a corrosion problem in the bottom section. Hence, the catalyst flow rate can only be changed slightly and it is usually kept at about 1% of the AC feed flow rate.20 3.5. Practical Degrees of Freedom. The identified dominant variables are column pressure, reflux ratio, temperature profile, feed ratio (feed rates), composition profile, and catalyst activity. The column pressure cannot be manipulated since it has to be maintained at around 1 atm. The temperature profile and the reflux ratio are not independently controllable since the reflux ratio affects the temperature profile. Similarly, the composition profile and the feed ratio are not independent. Thus, out of these six dominant variables, only three are independently controllable, which are the practical degrees of freedom. 3.6. Proposed Control Structure. To achieve the desired column temperature profile, it is sufficient to control the temperature at a single stage.13,18 The chosen tray is located in the upper part of the reactive section where a sensitive change in column temperature takes place due to a sharp change in the water fraction.20 Reflux ratio (or reflux rate), which strongly affects the stage temperature, is the manipulated variable chosen for this control. However, the real manipulated variable is the steam flow rate to the reboiler, which changes the vapor up-flow to compensate for the changes in liquid down-flow under a constant distillate flow rate in Figure 4. Aspen Plus simulations show that the top reaction temperature has good sensitivity with respect to the reflux ratios of 1.3-3.0 and does not have any output multiplicity in Figure 5a. Input multiplicities occur in the overall reaction conversions in Figure 5b. In other words, the same reaction conversion occurs for two different reflux ratios. However, the reaction conversion is above 99% in the given range of the reflux ratios and the input multiplicities do not significantly affect the reaction performance. The product specification of 99% MA is met when the reflux ratio is above 1.6 in Figure 5c.

Ind. Eng. Chem. Res., Vol. 43, No. 2, 2004 379

Figure 4. Partial control scheme for the methyl acetate reactive distillation system.

Figure 5. Steady simulation results using WILSON-NTH in Aspen Plus with reaction equilibrium data in Song et al..29 Reflux ratio versus (a) top reaction temperature, (b) overall reaction conversion, and (c) methyl acetate purity at the top, respectively.

The composition profile is controlled by the feed ratio of MT and AC. This feed ratio can be manipulated by changing either MT or AC flow rate. Aspen Plus simulations confirm that for a reactive distillation

column a small excess of AC (i.e., 5%) has a much stronger effect on increasing reaction conversion than a large excess of MT (>50%) in the reactive distillation column. Hence, in the case of poor reaction performance, excess AC can give much better compensation than excess MT. Thus, the AC flow is manipulated to control the composition profile in Figure 4. However, to achieve this improvement, we pay a price because it is very expensive to recover the excess AC due to the tangent pinch between W and AC. The catalyst activity is controlled by manipulating the SA flow rate. Different market demands of the product MA require different levels of the feed flow rate and the SA flow rate, but the control structure will remain the same as long as the equipment capacity can accommodate the production demands. The inventory scheme includes two liquid level controls on the overhead drum level and bottom level in addition to one vapor inventory (pressure) control. The top level is controlled by the distillate flow (D), the bottom level by the bottom product flow (B), and the pressure by a flow of inert N2. The control structure is completed by one constraint control of the bottom temperature to prevent corrosion due to SA. The manipulated variable for this task is the methanol feed flow rate because it strongly changes the bottom temperature without significantly affecting other variables in the process. The overall control scheme is summarized in Table 2. 3.7. Possible Sources of Instability. One important feature of the control structure proposed in Figure 4 is that the productivity (D, distillate flow rate) is maintained constant and the steam flow rate (reboiler duty) is manipulated to change the reflux ratio (or reflux rate). This is different from the control structure found in the literature20,21 where D is manipulated to change the reflux rate under a relatively constant reboiler duty. If D is manipulated to control the top temperature, it is not possible to ensure that the system meets the process specification (productivity, D). Also in this case, the

380

Ind. Eng. Chem. Res., Vol. 43, No. 2, 2004

Table 2. Control Scheme for Reactive Distillation Partial Control Structure dominant variables manipulated variables top reaction temperature reflux flow rate (or steam to the reboiler) column composition profile acetic acid feed flow rate catalyst activity sulfuric acid flow rate Inventory Control Structure pressure nitrogen flow to overhead drum overhead drum level distillate flow rate bottom level bottom byproduct flow rate Constraint Control Structure bottom temperature methanol feed flow rate Table 3. Limitations of Reactive Distillation with Solid Catalysts inside a Column items reaction and distillation temperature solid catalysts flow distribution inside catalytic trays or packings space velocity

requirements and limitations common temperature range active at the same common temperature range for a reasonable time good flow pattern required for high reaction conversion determined by distillation, unavailable to be adjusted for reaction

output multiplicities of the top reaction temperature, reaction conversion, and MA purity can occur as recently reported by Al-Arfaj and Luyben30 using pure reactants. For the case where a feed impurity such as n-propyl acetate is considered in the AC feed stream, these output multiplicities also arise in addition to input multiplicities.31 However, the proposed control structure in Figure 4 can achieve the required specifications and can avoid the occurrence of output multiplicities. 4. Widening the Applicability of Reactive Distillation The study of reactive distillation, by using concurrent design based on partial control, has allowed us to understand what are the limitations of the technology and how to change the design so that it can be applied to a wider range of industrial processes. In a previous section (3.2), we discussed several limitations of a reactive distillation column where reaction occurs inside the column. Table 3 summarizes these limitations. First of all, separation temperature and reaction temperature have to match and a catalyst must be active at the same conditions. A homogeneous catalyst is preferable to a heterogeneous one because the latter could cause flow maldistribution24 and frequent and costly shutdowns to regenerate and replace the spent catalyst. Finally, for a heterogeneous catalyst, space velocity cannot be freely adjusted since it depends on the separation requirements of internal flow rates (reflux rate). As there are upper and lower limits to the space velocity, these constrain the maximum amount of catalyst one can put into a column. The need to match both catalyst activity and temperature of the reaction to the distillation strongly limits the applicability of reactive distillation. 4.1. External Reactors Cascaded into a Column. All the above limitations can be avoided by going back to the design in Figure 3a, where external reactors are cascaded into a nonreactive distillation column. The central idea is the decoupling of the two tasks (reaction and distillation), which gives an increment of available independent degrees of freedom. This increment allows

Figure 6. Schematics of (a) conventional reactive tray and (b) simulated external reactor.

us to independently control both reaction and separation temperatures. Furthermore, with external reactors, complex design of column internals can be eliminated since we can use traditional reaction engineering techniques for designing the external reactors. Finally, space velocity can be adjusted by simply changing the size of external reactors. This decoupling results in a wider applicability of the reactive separation technology when reaction and separation conditions are different or/and heterogeneous catalysts are needed. The external ()side) reactor concept in reactive distillation was introduced in the early 1980s by Schoenmakers and Buehler.32 Since reaction can be controlled independently of distillation, many applications or derivatives of the side-reactor distillation can be envisioned for heterogeneous catalytic systems. Despite its potential there is little research effort on the design and control of this side-reactor technique.33 One commercialization is the tert-amyl methyl ether (TAME) production system.34 Recently, this idea was also applied to the olefin dimerization process.35 4.2. Comparison of Performance between Reactive Distillation and External Reactor Column. The conventional reactive tray and the simulated side reactor are sketched in Figure 6. The liquid and vapor streams are fed to the side reactor and recycled back to the column after being reacted. They are entirely withdrawn from the same positions and return to the same trays as if the reactor were located inside the column. So the internal vapor and liquid flow rates of the side-reactor case are basically identical to those of the current reactive distillation column. Heat exchangers before and after the reactors allow independent control of reaction temperature. Further, it is possible to pack heterogeneous catalysts inside the side reactors and build bigger reactors to have larger space velocities. Replacement or regeneration of spent catalysts can be performed without shutting down the entire column. The process performance is compared between the side-reactor column and the reactive distillation column

Ind. Eng. Chem. Res., Vol. 43, No. 2, 2004 381

Figure 7. Comparison between a reactive distillation column with 39 reactive trays and conventional distillation column with four side reactors (both with 60 total trays and assuming 100% equilibrium conversion).

by using the methyl acetate example. Here, several possible scenarios for heterogeneous catalysis are considered in the Aspen Plus simulations of the two cases. It is assumed in the simulations that all feed streams to the reactive column are saturated liquid. The simulation results in Figure 7 confirm that four side reactors are sufficient to practically obtain the same performance as a reactive distillation column with 39 reactive trays.

The total number of trays is 60 for the two cases. Both can produce on-spec methyl acetate within the acceptable range of reflux ratios (1.7-2.7) under the assumption of 100% reaction equilibrium. The side-reactor case also allows us to overcome some problems encountered in the internal catalytic-tray case, especially with heterogeneous catalysts. The two major problems refer to poor reaction conversion and low tray separation efficiency due to the short residence time and flow maldistribution. Low reaction conversion has been considered as 80% and 60% of reaction equilibrium conversion for internal-catalytic trays or catalytic packing. However, 90% equilibrium conversion is assumed for the side reactors since reaction temperature could be optimized and larger catalyst hold-ups can be provided without being constrained by separation requirements. The internal-catalytic case with 80% equilibrium conversion and the side-reactor case give the product purity slightly off the required specification as shown in Figure 8. But the 60% equilibrium conversion in the internal-catalytic case leads to much lower MA purity than its required specification. Figure 9 refers to the same three cases when a 5% excess of AC is provided to meet the product specification. The case with 60% equilibrium still does not meet the specs while the other two cases do.

Figure 8. Impact of low reaction performance on the product specs with 0% acetic acid excess.

Figure 9. Impact of low reaction performance on the product specs with 5% acetic acid excess.

382

Ind. Eng. Chem. Res., Vol. 43, No. 2, 2004

Figure 10. Impact of low separation efficiency on the product specs with 5% acetic acid excess.

Lower separation efficiency can arise due to the flow maldistribution through the column internals packed with solid catalysts. This situation is considered by reducing the total number of theoretical stages from 60 to 45, of which 24 are reactive trays. This means that the separation efficiency is about 60% (24 out of 39 stages) for the catalytic trays. For the side-reactor case, the total stage number of the distillation column is assumed not to change since there is no catalyst inside the column that can cause a reduction in separation efficiency. Figure 10 shows that on-spec MA can be produced in both cases for 80% and 90% equilibrium conversion when 5% acetic acid excess is used. Furthermore, the side-reactor case has a wider operating range of reflux ratios than the internal catalytic case with respect to the required product specification in Figure 10. Therefore, from these simulation results in Figures 7-10, the side-reactor column achieves almost the same performance as the reactive distillation column and has even better performance when lower reaction and separation efficiency are considered. 5. Discussions and Conclusions We have applied concurrent design method, based on partial control, to reactive distillation and have derived a countercurrent scheme (where side reactors are cascaded into a conventional distillation column) to enlarge the applicability of reactive distillation. By benchmarking Eastman’s methyl acetate production system, we arrive at this countercurrent side-reactor column design suitable for most of the reaction systems where reaction and distillation conditions are mismatched, or for which there are only heterogeneous catalysts available. We showed that this countercurrent scheme in reactive distillation circumvents separation limitations due to azeotropes and reaction equilibrium constraints. Furthermore, this scheme also reduces significant recycle flow of inerts in the conventional reactor-separator-recycle flowsheet as already shown in the reactive distillation for ether production (MTBE, ETBE, and TAME).36 The concurrent design of reactive distillation enabled us to understand what are the limitations in current applications, why its use is not prevalent yet, and what structural changes are needed. In fact, a process must

have a catalyst that is active at a temperature that matches separation temperature under an operating pressure. The pressure, in most cases, has such a strong impact on the boiling points and relative volatility that cannot be manipulated to change the separation temperature in order to match the reaction requirements. High catalyst activity is essential for avoiding very large hold-ups that would be necessary to achieve required reaction conversion. When the catalyst is heterogeneous, we also found that other problems would arise. First of all, a solid packed in a column requires complex tray internal designa to reduce flow maldistribution. Then, more trays have to be built to compensate for the deterioration of separation efficiency in the reactive section. Furthermore, each reactive tray has to be designed bigger than a conventional tray to provide large catalyst hold-ups and guarantee sufficiently long residence times. Neither the internal flow rate nor the catalyst hold-up is available as manipulated variables to adjust the space velocity. Thus, we lose a degree of freedom. Finally, it is not possible to perform on-line regeneration and replacement of spent catalysts. The design of external reactors cascaded into a distillation column allows us to overcome all the limitations discussed above, especially when using heterogeneous catalysts. In particular, reaction and separation temperatures can be controlled independently, and thus regain a degree of freedom. Reaction temperature can be then optimized in order to maximize catalyst activity. Continuous catalyst regeneration is possible since we can cyclically shut down one reactor while operating the rest of them. Furthermore, we can build reactors with large volumes; this allows us to adjust the space velocity independently of separation, by providing more catalyst into the reactors. This design has great potential for extending the applicability of reactive distillation. However, more research effort should be made for the feasibility studies of distillation columns cascaded with side reactors. Acknowledgment The authors are grateful to Profs. Rinard and Shinnar at the City College for their valuable comments on this paper.

Ind. Eng. Chem. Res., Vol. 43, No. 2, 2004 383

Notation D ) Distillate flow rate Ln ) molar liquid flow rate at stage n Tn ) temperature at stage n Vn ) molar vapor flow rate at stage n Abbreviations AC ) acetic acid DC ) distillation column HX ) heat exchanger MA ) methyl acetate MT ) methanol R ) reactor SA) sulfuric acid W ) water

Literature Cited (1) Arbel, A.; Huang, Z.; Rinard, I. H.; Shinnar R.; Sapre A. V. Dynamics and Control of Fluidized Catalytic Crackers. 1. Modeling of the Current Generation of FCC’s. Ind. Eng. Chem. Res. 1995, 34, 1228. (2) Arbel, A.; Rinard, I. H.; Shinnar R.; Sapre A. V. Dynamics and Control of Fluidized Catalytic Crackers. 2. Multiple Steady State and Instabilities. Ind. Eng. Chem. Res. 1995, 34, 3014. (3) Arbel, A.; Rinard, I. H.; Shinnar R. Dynamics and Control of Fluidized Catalytic Crackers. 3. Designing the Control System: Choice of Manipulated and Measured Variables for Partial Control. Ind. Eng. Chem. Res. 1996, 35, 2215. (4) Arbel, A.; Rinard, I. H.; Shinnar R. Dynamics and Control of Fluidized Catalytic Crackers. 4. The Impact of Design on Partial Control. Ind. Eng. Chem. Res. 1997, 36, 747. (5) Shinnar, R.; Rinard, I. H.; Dainson B. Partial Control. 5. A Systematic approach to the Concurrent Design and Scale-up of Complex Processes: The Role of Control System Design in Compensating for Significant Model Uncertainties. Ind. Eng. Chem. Res. 2000, 39, 103. (6) Kothare, M. V.; Shinnar, R.; Rinard, I.; Morari, M. On Defining the Partial Control Problem: Concept and Examples. AIChE J. 2000, 46, 2456. (7) Buckley, P. A. Techniques of Process Control; Wiley: New York, 1964. (8) Hougen, J. O. Measurements and Control Applications; Instrument Soc. of America, 1979. (9) Morari, M.; Arkun, Y.; Stephanopolous, G. Studies in the Synthesis of Control Structures for Chemical Processes: I. Formulation of the Problem. Process Decomposition and the Classification of the Control Tasks. Analysis of the Optimizing Control Structures. AIChE J. 1980, 26, 220. (10) Govind, R.; Powers, G. J. Control System Synthesis Strategies. AIChE J. 1982, 28, 60. (11) Douglas, J. M. Conceptual Design of Chemical Processes; McGraw-Hill: New York, 1988. (12) Georgiou, A.; Floudas, C. A. Structural Analysis and Synthesis of Feasible Control Systems. Chem. Eng. Res. Des. 1989, 67, 600. (13) Luyben, W. L.; Tyre´us, B. D.; Luyben, M. L. Plantwide Process Control; McGraw-Hill: New York, 1998. (14) Zheng, A. R.; Mahajanam, R. V.; Douglas, J. M. Hierarchical Procedure for Plantwide Control System Synthesis. AIChE J. 1999, 45, 1255. (15) Rinard, I. H.; Downs, J. J. Plant Wide Control. A Review and Critique. AIChE Annual Meeting, 1992 (Spring).

(16) Larsson, T.; Skogestad, S. Plantwide ControlsA Review and a New Design Technique. Model. Identif. Control 2000, 21, 209. (17) Shinnar, R. Chemical Reactor Modeling for Purposes of Controller Design. Chem. Eng. Commun. 1981, 9, 73. (18) Tyre´us, B. D. Dominant Variables for Partial Control. 1. A Thermodynamic Method for Their Identification. Ind. Eng. Chem. Res. 1999, 38, 1432. (19) Tyre´us, B. D. Dominant Variables for Partial Control. 2. Application to the Tennessee Eastman Challenge Process. Ind. Eng. Chem. Res. 1999, 38, 1444. (20) Agreda, V. H.; Partin, L. R. Reactive Distillation Process for The Production of Methyl Acetate. U.S. Patent 4,435,595, Mar 6, 1984. (21) Agreda, V. H.; Partin, L. R.; Heise, W. H. High Purity Methyl Acetate via Reactive Distillation. Chem. Eng. Prog. 1990, 86 (2), 40. (22) Siirola, J. J. An Industrial Perspective on Process Synthesis. AIChE Symp. Ser. 1995, 304, 222. (23) Doherty, M. F.; Buzad, G. Reactive Distillation by Design. Trans. Inst. Chem. Eng. 1992, 70, 448. (24) Taylor, R.; Krishna, R. Modelling Reactive Distillation. Chem. Eng. Sci. 2000, 55, 5183. (25) Malone, M. F.; Doherty, M. F. Reactive Distillation. Ind. Eng. Chem. Res. 2000, 39, 3953. (26) Lee, J. W.; Westerberg, A. W. Graphical Design Applied to the MTBE and Methyl Acetate Reactive Distillation Processes. AIChE J. 2001, 47, 1333. (27) Bessling, B.; Schembecker, G.; Shimmrock, K. H. Design of Processes with Reactive Distillation Line Diagrams. Ind. Eng. Chem. Res. 1997, 36, 3032. (28) Lee, J. W.; Bru¨ggemann, S.; Westerberg, A. W. Visualization of the Ethyl Acetate Reactive Distillation System. Proceedings of ECCE 3, Nu¨rnberg, June 2001. (29) Song, W.; Venimadhavan, G.; Manning, J. M.; Malone, M. F.; Doherty, M. F. Measurement of Residue Curve Maps and Heterogeneous Kinetics in Methyl Acetate Synthesis. Ind. Eng. Chem. Res. 1998, 37, 1917. (30) Al-Arfaj, M. A.; Luyben, W. L. Comparative Control Study of Ideal and Methyl Acetate Reactive Distillation. Chem. Eng. Sci. 2002, 57, 5039. (31) Citro, F.; Lee, J. W. Partial Control and Concurrent Design of Integrated Reaction and Distillation. AIChE Annual Meeting, 2003 (Spring). (32) Schoenmakers, H. G.; Buehler, W. K. Distillation Column with External ReactorssAn Alternative to the Reaction Column. Ger. Chem. Eng. 1982, 5, 292. (33) Jakobsson, K.; Pyhlahti, A.; Pakkanen, S.; Keskinen, K.; Aittamaa, J. Modelling of a Side Reactor Configuration Combining Reaction and Distillation. Chem. Eng. Sci. 2002, 57, 1521. (34) Jrvelin, H.; Tamminen, E.; Ewy, G. NexTAME Process Operating Experiences from a Commercial Unit and Some Economical Considerations. NPRA Annual Meeting, San Antonio, 1996. (35) Sloan, H. D.; Birkhoff, R.; Gilbert, M. F.; Nurminen, M.; Pyhlahti, A. Isooctane Production from C4’s as an Alternaive to MTBE. NPRA Annual Meeting, San Antonio, 2000. (36) DeGarmo, J. L.; Parulekar, V. N.; Pinjala, V. Consider Reactive Distillation Chem. Eng. Prog. 1992, 88 (3), 43.

Received for review February 20, 2003 Revised manuscript received May 22, 2003 Accepted May 29, 2003 IE030172F