Feasible Separation Modes for Various Reactive Distillation Systems

Department of Chemical Engineering, Kyoto University, Kyoto 606-8501, Japan, ... Fine Chemical Technology, 117571 Vernadskogo Prospect, 86, Moscow, Ru...
0 downloads 0 Views 95KB Size
4060

Ind. Eng. Chem. Res. 1999, 38, 4060-4067

Feasible Separation Modes for Various Reactive Distillation Systems S. Giessler,*,† R. Y. Danilov,‡ R. Y. Pisarenko,‡ L. A. Serafimov,‡ S. Hasebe,† and I. Hashimoto† Department of Chemical Engineering, Kyoto University, Kyoto 606-8501, Japan, and Lomonosov Academy of Fine Chemical Technology, 117571 Vernadskogo Prospect, 86, Moscow, Russia

The feasibility of steady states of continuous reactive distillation processes is discussed by applying the theory of static analysis (SA). The generality of this approach is proved by applying several industrial examples, all of which have different reaction schemes and thermodynamical properties. For each example process, the entire feed composition region is divided into several subregions, each of which has similar characteristics for the product composition and the column structure. The information derived by this analysis can be used effectively for the selection of the desirable feed composition and column configurations. The results of various examples indicate that the SA is a very convenient tool that provides an answer to the question of feasibility and provides hints for an early stage of design. 1. Introduction

Table 1. Examined Types of Reactions

During the past decade, there has been a rapid increase in the interest in reactive distillation (RD), which has the potential for separating liquid mixtures involving azeotropes or isomers. RD is applied to esterification processes,1 the separation of isomers,2 and the production of “antiknock” enhancers such as MTBE and TAME.3,4 With the RD process, it is possible to overcome limitations caused by the thermodynamic chemical equilibrium. Therefore, compared with a simple combination of one reaction unit and one distillation unit, there exists the possibility of saving energy and reducing the environmental load. Following research in the academic sector, various successful commercial RD processes have been developed. An overview of the existing patents and literature about these processes is given by Ruiz et al.5 For an RD process, it is very difficult to estimate the achievable product regions from the feed composition and the vapor-liquid equilibrium (VLE) of the mixture. Thus, it is important to evaluate the feasibility of the given structure at an early stage of the design. For the design of nonideal systems, the state of the art has been described by Fidkowski et al.6 The approaches taken in such studies are based on the topological analysis of residue curves. Barbosa7 and Buzad8 also proposed design methods for RD that are based on the analysis of residue curve maps. An overview of the research of this group is given by Doherty and Buzad.9 On the basis of the above methods of Doherty and others and the application of heuristic rules and numeric routines, a design method that takes into account important elements of an overall design strategy (e.g., types of phase equilibrium, degrees of freedom, various reaction types and number of components) has recently been developed by Bessling et al.10 In addition, Nisoli et al.11 recently presented a more global method that * E-mail: [email protected]. Tel: +81-75-7534927. Fax: +81-75-752-9639. † Kyoto University. ‡ Lomonosov Academy of Fine Chemical Technology.

reaction scheme

no. of components

type of phase eq.

A+BSC A+BSC A+BSC+D A + B S C + Inert

3 3 4 4

VLE VLLE VLE VLE

combines a geometric approach to reactor synthesis with a geometric approach to the feasibility analysis of separations in order to find possible process alternative systems. Nisoli et al.11 used a parameter continuation method for the Damkoehler number and executed the simulations for various ratios of the liquid and vapor flows in the column. With a thermodynamic-topological analysis as the first step of a design procedure, a large number of configurations of combined processes can easily be examined and reduced to a smaller number of operation sequences. An approach that uses a thermodynamictopological analysis for the design of RD systems whereby the statics of an RD process are analyzed has been introduced by the group of Serafimov.12-14 The aim of this work is to check the performance of a systematic method for the determination of technically feasible product regions of various RD systems. The method used in this research is based on the static analysis (SA) developed by Pisarenko et al.15-17 and Danilov18 and is used to analyze the feasibility of RD processes, which can later be connected with an overall design strategy. In the SA for a fixed feed composition, the steady state of maximum feasible conversion is determined. In addition to a proof of feasibility, initial hints for the design of the RD column, such as the length and location of the reactive zone, are derived. These results can then be used to determine a column configuration. Table 1 gives an overview of the systems discussed in this paper. By changing the following factors, the generality of the method has been analyzed: (a) reaction scheme and number of components, (b) type of phase equilibrium, (c) number of azeotropes, and (d) presence of inerts. After first outlining the basics of the SA, the performance of the procedure is explained. The results of the

10.1021/ie9900162 CCC: $18.00 © 1999 American Chemical Society Published on Web 08/27/1999

Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999 4061

Figure 1. Distillation diagram of a three-component reactive system A + B S C.

SA are discussed and compared to other research works using five industrial applications. 2. Theory of Static Analysis SA, developed by Serafimov et al.,12 is based on a thermodynamic-topological analysis of distillation diagrams. The advantage of SA is that it represents a general method that does not depend on the concrete column structure. The principal assumptions in SA are as follows: (i) The vapor and liquid flow rates in the column are infinite. (ii) The capacity of the reaction part in the column is large enough to carry out a given conversion rate, and the reaction part is located at a certain place in the column. (iii) The plant is operated at steady state, and theoretical stages are chosen. (iv) One reversible equilibrium reaction is considered. By introducing the assumption of infinite flow rate, the composition change, caused by the reaction on each tray, can be neglected. Furthermore, it is possible to regard the RD as successive operations of reaction and distillation. First, the feed material is converted to a pseudoinitial mixture, X*, by assuming the reaction has occurred to a certain extent. The mixture is then separated by a conventional distillation column. From the assumptions mentioned above, the liquid composition on a tray can be regarded as the same as that of the vapor flowing into the same tray. Thus, the composition profile in the column can easily be calculated by using the usual distillation lines. An example of the distillation line is shown in Figure 1. In this example, the reaction scheme is A + B S C. For a given feed composition XF, the composition of a pseudoinitial mixture, X*, is calculated by assuming the conversion rate of a reagent. The line of chemical interaction (LCI) is obtained by changing the conversion rate between 0 and 100%. The boundary between the forward and reverse reaction region is called the chemical equilibrium manifold (CEM). When a composition of a pseudoinitial mixture X* is given, this steady state is feasible if the corresponding distillation line satisfies the following two conditions: (I) The composition of a pseudoinitial mixture X* and the product compositions satisfy the material balance. (II) A part of the distillation line lies inside the forward reaction region. It is possible to calculate the number of theoretical stages from the distillation line and the number of reactive stages by using the intersection of the distillation line and the CEM. There are many pairs of pseudoinitial mixtures and product compositions that satisfy the material balance. Therefore, in the SA, one of the product compositions is fixed in advance. For the structure, which is called hereafter “direct separation,” the lightest composition attainable from X* is fixed as

a reference point, and it is assumed that the top composition is the same as that of the reference point. On the other hand, for “indirect separation,” the heaviest composition is fixed as a reference point and is assumed to be equal to the bottom composition. When the composition of one product is fixed, the composition of the other product is restricted by the material balance; i.e., two products and the pseudoinitial mixture must lie on the same line in the distillation diagram, as shown in Figure 1. This line is hereafter called a “line of mass balance” (LMB). It should be noted that the LMB is divided by the pseudoinitial mixture into two segments at a certain ratio, which is the same as that of the amount of distillate and the bottom product (D/B). In SA, when a product is fixed as a reference point, all of the available compositions on the LMB are examined to evaluate whether a given degree of conversion is feasible or not. If in the procedure a given degree of conversion is found to be feasible (infeasible), the calculation of the distillation line is repeated for a pseudoinitial mixture of higher (lower) conversion until the maximum conversion is found. By increasing the conversion rate of a reagent, the reference point for the direct (or indirect) separation might be changed. In this research, only the maximal possible conversion of a reagent is derived for various feed compositions, and the characteristics of each RD system is investigated by classifying the feed and product compositions. The condition of the maximal possible conversion is hereafter called a “limiting steady state” (LSS), and the distillation line for the LSS is hereafter called a “limiting path”. The features of an LSS are described by Pisarenko et al.14-17 The proposed method, using minimal information about the physicochemical properties of the reaction mixture, calculates the maximum degree of conversion, the reactive zone location and its quantitative length. The data used are (i) feed composition, (ii) phase equilibrium model parameters, (iii) chemical equilibrium model parameters, and (iv) stoichiometry of the reaction. The steps for evaluating the feasibility of a steady state of the process referring to a certain degree of conversion are described in detail by Giessler et al.19 3. Case Studies 3.1. Production of Acetic Acid. In the case of the production of acetic acid, the reaction scheme is as follows:

acetic anhydride + water S acetic acid (bp 139 °C) (bp 100 °C) (bp 118 °C) By executing SA for various initial feed compositions, it becomes clear that the whole feed composition region can be divided into subregions by 4 characteristic points: XF1, XF2, XF3, and XF4 in Figure 2. Here, XF1 is the starting point for the LCI that ends at AzAA-AcAn, XF2 is the starting point of the LCI that intersects with the CEM and the distillation boundary, XF3 is the point of a stoichiometric input, and XF4 is the AzAcAn-W (37 mol % acetic anhydride). For each region, the column configuration is the same for all initial conditions. That is, within each region, only the number of trays and the flows change, but the structure remains unchanged. The

4062

Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999

Figure 3. Result of the indirect separation for the AA production.

Figure 2. Result of the direct separation for the AA production. Table 2. Product Regions for the Production of Acetic Acid W comp. in feed X < XF1 XF1 XF1 < X < XF2 XF2 XF2 < X < XF3 XF3 XF3 < X < XF4 XF4 X > XF4 X < XF1 XF1 XF1 < X < XF2 XF2 XF2 < X < XF3 XF3 XF3 < X < XF4 XF4 X > XF4

type of sep. direct direct direct direct direct direct direct direct direct indirect indirect indirect indirect indirect indirect indirect indirect indirect

top comp. AzAcAn-W AzAcAn-W AzAcAn-W AzAcAn-W AzAcAn-W AzAcAn-W AzAcAn-W AzAcAn-W AzAcAn-W mixture of all mixture of all mixture of all mixture of all mixture of all mixture of all W W W

bottom comp. AcAn/AA AzAA-AcAn AcAn/AA AcAn/AA AcAn/AA AA AA/W AA/W AA/W AcAn AcAn AcAn AcAn AA AA AA AA AA

top and bottom compositions in each region for the direct and indirect separations are listed in Table 2. Next, the case of a stoichiometric feed, which is a commonly used input for an RD process, is discussed in more detail. For both direct and indirect separations, the results of SA show that the acetic acid is withdrawn from the bottom and that there is little product withdrawal from the top, i.e., most of the acetic anhydride and water are recycled. The simulation result shows that the reactive zone consists of about seven stages and is located at the top of the column. The column consists of 24 stages for the direct separation and 15 stages for the indirect separation. The results for the direct and indirect separations are shown in Figures 2 and 3. For the direct separation, the AzAcAn-W is the top product. At the LSS, the pseudoinitial composition corresponds to pure AA. For this condition, the limiting path is AA-AzAA-AcAn-D. For the indirect separation, pure AA is the bottom product, and the LMB and the LCI take the same line. For this condition, the limiting path is expressed by AAAzAA-AcAn-D. For the indirect separation, the feasibility of the operation is strongly influenced by the chemical equilibrium constant. If point D in Figure 3 does not belong to the forward reaction region, this operation is infeasible. In RD, it is sometimes unnecessary to achieve a 100% conversion. Even for such cases, it is possible to estimate the feasibility of the operation and the top and bottom compositions. For example, pure AcAn product can be obtained as the bottom product as long as the pseudoinitial mixture lies within the triangle AzAcAn-W-

AzAA-AcAn-AcAn. The information obtained by changing the extent of the reaction is also very useful for further synthesis. 3.2. Production of Cumene. The reactions of olefins with aromatics are well-established from a commercial viewpoint. The reaction of benzene and propylene producing cumene is the most important among such reactions. During the reaction, cumene itself reacts with propylene and forms diisopropylbenzene, which also reacts with benzene to form more cumene in a transalkylation step. Recycling of these heavy byproducts to the reactor results in a higher yield of cumene. The advantage of the combined process is that benzene can be kept in the reaction zone to suppress the formation of byproducts. The reaction scheme at 513 kPa is as follows:

benzene + propylene S cumene (bp 80 °C) (bp 47 °C) (bp 152 °C) The results of SA for the direct and indirect separations, using a stoichiometric feed composition, are shown in Table 3. The equilibrium constant, estimated from thermodynamic data, is 5.0. In the case of the direct separation in Table 3, it is not possible to reach 100% conversion since for the steady state only one point of the distillation line lies on the CEM. Thus, an infinite amount of catalyst would be necessary in order to achieve this extent of conversion. As this is technically not feasible, the conversion rate for an actual system will be slightly smaller than that for the total conversion. The results of the indirect separation show that a part of the distillation line for total conversion belongs to the forward reaction region. Therefore, the steady state can be achieved by using a finite volume of catalyst. In both the direct and indirect separation, as specified above, cumene will be withdrawn as the bottom product. The reactive zone will be placed at the top of the column from the first to the second stage (stages are counted from the top of the column), and the column is optimally operated at total reflux (D/B ) 0). The number of ideal stages varies from 3 to 11, depending on the separation conditions and the column structure. The results of SA coincide with those of Shoemaker and Jones20 in that the product always consists of less than 2 wt % benzene. 3.3. Production of Ethylene Glycol. The production of ethylene glycol by the reaction of ethylene oxide and water is a commercial example of a large-scale parallel reaction. Corrigan and Miller21 claimed that one of the main problems lies in maximization of the amount of monoethylene glycol while reducing the byproducts of diethylene glycol and higher glycols. In this system, the liquid phase is split into a water and an EO phase. Therefore, VLLE must be employed for calculating the

Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999 4063 Table 3. Cumene Production for a Stoichiometric Feed Composition composition of products distillate

bottom

type of sep.

conv. [%] B

conv. [%] P

total stages

reac. zone

D/B [mol/mol]

B

P

C

B

P

C

direct indirect

99.79 100.0

99.79 100.0

3 10

1:1 1:2

0 0

0 0.5

1.0 0.5

0 0

0.001 0

0 0

0.999 1.0

physicochemical data. The reaction scheme for this process is as follows:

ethylene oxide + water S ethylene glycol (bp 10 °C) (bp 100 °C) (bp 207 °C) The patent of Parker22 shows that a much higher selectivity can be achieved by adopting an RD column. Parker reports that the more volatile components of the feed are immediately separated at the feed plate so that a large amount of unreacted water is left on the feed plate. The ethylene glycol formed, having a low volatility, passes down the column and is removed from the reaction zone. This both prevents further reaction and suppresses the parallel reaction. Therefore, the generation of byproducts is not taken into account in this paper. As this process consists of highly exothermic reactions, temperature-dependent reaction rates and large volatility differences between the components result in multiplicities. This system is a good example of applying SA, which is capable of locating and analyzing steady-state multiplicity because steady states between 0 and 100% conversion are thoroughly investigated during the analysis. Ciric and Miao23 analyzed this process by a homotopy method and found a very cost-effective column. The column consists of 10 trays and has no distillate product, and the reactive zone at the top consists mostly of water. In our paper, the same column pressure as that reported by Dudukovic24 is taken (100 kPa). The concentration profile achieved by Dudukovic, using a homotopy analysis, is similar to that obtained by SA. In both cases, the ethylene oxide concentration shows a sudden change in the upper middle part of the column. For a stoichiometric feed, the number of stages becomes 12 for the indirect separation and 6 for the direct separation. In both cases, trays 1 and 2 are reactive. The conversion rate for the LSS is 99.8%. Ethylene glycol is always withdrawn as a bottom product. The characteristics of the results are almost identical with those obtained for cumene production (see section 3.2). 3.4. Production of Ethyl Acetate or Methyl Acetate. Transesterifications are often employed in RD, as they are equilibrium-limited reactions that proceed to completion only if one component is continuously removed from the reaction mixture. Usually, these reactions are carried out under mild conditions, which makes them suitable for RD. In industry, esterifications and transesterifications play a major role in reducing the boiling point of esters by exchanging a long-chain alkyl group. The reaction producing ethyl acetate is slow and mildly exothermic when operated at temperatures between 76 and 117 °C and standard pressure. The reaction scheme of the process is

ethanol + acetic acid S (bp 78.4 °C) (bp 118 °C) ethyl acetate + water (bp 72.2 °C) (bp 100 °C)

Figure 4. Direct separation for the production of ethyl acetate (infeasible).

The present multicomponent system has four azeotropes. The conditions of the azeotropes and the physicochemical data at 100 kPa are listed in the Appendix. The chemical equilibrium constant at 70 °C has been measured to be 10.6.18 When a feed composition XF is fixed, the first step of the feasibility analysis is to calculate the LCI and to choose a pseudoinitial mixture. The composition diagram for the four-component system is shown in Figure 4. In this figure, X* shows the point of 100% of conversion of ethanol. In the direct separation of this system, the distillation line starting from any pseudoinitial mixture converges in the three-component azeotrope AzE-EA-W. Thus, for the case of the direct separation, the top product is AzE-EA-W. The LMB connecting the fixed top product and a pseudoinitial mixture, X*, is shown in Figure 4. The distillation line starting from X* is also shown in the figure. In this case, the distillation line lies in the reverse reaction area. Thus, this steady state is infeasible. This example shows that even for the cases where the concentration of ethanol in the feed is lower than 50 mol %, a 100% conversion of ethanol cannot be achieved. It is found that when the LMB, from the reference point AzE-EA-W to the bottom product composition, reaches the composition plane of AA-EA-W, the corresponding degree of conversion is infeasible. In other words, when the LMB, from AzE-EA-W to the bottom product composition, reaches the composition plane of AA-E-W, the corresponding degree of conversion is feasible. The LSS corresponds to the case where the LMB, from AzE-EA-W to the bottom composition, reaches B1, which is the edge of AA-W. The pseudoinitial mixture for this case is X**. For this case, two limiting paths can be drawn: B1-W-AzEA-W-AzE-EA-W and B1-W-AzE-W-AzE-EA-W as shown in Figure 5. As both paths are on the edge of the forward reaction region, they thus lie on the boundary between being feasible and not feasible. Therefore, a steady state with a conversion rate slightly less than the LSS will actually be chosen (bottom product must lie on the plane AAE-W). The distillation line for a finite number of trays

4064

Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999

Figure 5. Direct separation for the production of ethyl acetate (feasible). Table 4. Product Regions of the Esterification of Ethanol and Acetic Acid feed comp. of AA

kind of sep

top product

bottom product

0-0.473 0.474 0.475-0.49 0.5-1.0 0-0.49 0.5-1.0

direct direct direct direct indirect indirect

AzE-EA-W AzE-EA-W AzE-EA-W AzE-EA-W E/EA/W (trace of AA) EA/W (trace E, AA)

E/W (trace of AA) W AA/W (trace E, EA) AA/W (trace of E, EA) AA AA

Figure 7. Result of the direct separation for butyl acetate.

As the production process of methyl acetate is similar to that of ethyl acetate, only the production of ethyl acetate has been discussed in this section. 3.5. Production of Butyl Acetate. Among the tested systems, the production of butyl acetate is the most complex, as it has six azeotropes. These are listed in the Appendix. This esterification differs from the esterifications mentioned before in the sense that the desired product, butyl acetate, is the component with the highest boiling point and that a phase splitting, which has to be taken into account in the calculation of the VLE data, occurs. The esterification of butanol is executed at quite a low pressure, as the catalyst would be destroyed at a higher pressure. The reaction scheme at 32 kPa is

Figure 6. Indirect separation for a feed with a little excess of AA.

butanol + acetic acid S (bp 89.41 °C) (bp 85.6 °C) water + butyl acetate (bp 71.6 °C) (bp 91.6 °C)

is the curved line shown in Figure 5. A large part of this line belongs to the forward reaction region. For the case where the concentration of the ethanol in the feed is higher than 52.6 mol %, the acetic acid can be totally converted, and the bottom product is a mixture of ethanol and water. All possible composition regions are listed in Table 4. The regions listed characterize the typical column structures achieved by executing the procedure of SA for the entire feed composition region. The influence of the azeotropes that make the separation more difficult can easily be seen, as all the achieved top products are azeotropes. In the indirect separation, all distillation lines converge in pure AA, as this is the highest boiling reference point. A result for the indirect separation with a feed composition XF consisting of more than 50 mol % AA is shown in Figure 6. As the distillation line for 100% conversion lies in the reverse reaction region, this state is not feasible. In other words, ethanol cannot be exhausted totally by the reaction. A state with a little less conversion, where D1 is the distillate and B1 is the bottom product, is proved to be feasible. The limiting path is the dashed line in Figure 6. For a feed composition containing less than 50 mol % AA, a 100% conversion of AA is feasible. In this case, the top product is always a mixture of E, EA and W.

In Figure 7, the result of the direct separation for a stoichiometric feed XF is shown. For the pseudoinitial mixture with total conversion X*, the resulting distillation line X*-AzBA-W-AzB-BA-W belongs to the reverse reaction region. When the bottom product lies on the edge BA-AA, e.g., in the case of a pseudoinitial mixture X**, one point of distillation line lies on the CEM. Thus, this state is feasible but only for a reactive zone of infinite volume. The limiting path for this condition is B1**-AA-W-AzBA-W-AzB-BA-W. As the distillation line for X** lies on the edge of the CEM, a pseudoinitial mixture with slightly less conversion than X** will actually be chosen. The entire feed region is divided into four subregions by three characteristic points: the stoichiometric feed, the AzAA-B, and XFT in Figure 8. Here, XFT is the starting point for the LCI that ends at the distillation boundary. By use of the indirect separation, pure BA can be obtained as bottom product. The result for an initial feed of 70 mol % AA is shown in Figure 8. Here, two distillation lines for the pseudoinitial mixture with total conversion of butanol X* can be drawn: (1) D1AA-BA and (2) D1-AA-AzAA-B-BA-BA. As the first line is not lying in the forward reaction region, the limiting path is the second one. In this example, the composition simplex is divided into two regions by the gray distillation boundary in Figure 8. It should be noted that if the desired product is not BA but is another

Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999 4065

Figure 8. Result of the indirect separation for butyl acetate with 70% AA in the feed.

material, both distillation regions should be examined. In this case, for low conversion rates, the AzAA-B is withdrawn as the bottom product, whereas BA is withdrawn for high conversion rates. By executing SA for various feed compositions, it becomes clear that in the case of the indirect separation, pure BA can be obtained as the bottom product for the feed between the stoichiometric feed and XFT in Figure 8. For feed concentrations between XFT and the pure B, the bottom product will always be AzAA-B. If the LMB from the lowest boiling point AzB-BA-W reaches pure BA, it is also possible to obtain pure BA in the direct separation. This is the case with a feed composition having a slight excess of B. The difference between the esterification of ethanol and that of butanol is the order of boiling points of materials. For the esterification of ethanol, the boiling points of the azeotropes are slightly lower than that of the desired product. For this reason, even in the direct separation, pure EA could never be obtained. The five different kinds of separation show that the results of the feasibility study provide useful information for a more detailed synthesis. For example, in the production of BA, the feed regions where pure BA can be produced are calculated. Because of the knowledge of the achievable compositions and the extent of the reaction, hints for a decision to use an additional RD column or another type of separator are provided. 4. Conclusion Static analysis of combined processes has been successfully applied to the first step of the design of a variety of commercial processes for producing acetic acid, esters, and other organic products. On the basis of the results described in this paper and in the previous paper19 about the MTBE production, which considers inerts, it has been verified that the procedure is applicable for mixtures, with or without inerts, as well as mixtures that have different reaction schemes and different phase equilibria. The SA adopted in this research provides not only knowledge for predicting the limiting potentials of single RD columns but also valuable information for feasible column configurations. As the feasible product composition regions can easily be determined, it is also easy to check whether a desired extent of reaction can be achieved for a given feed composition. When the mixture forms azeotropes, it is very difficult to focus the feasibility of reactive distillation processes. It has been verified,

with example problems, that SA can easily be applied to processes with azeotropes. One of the dominant characteristics of the present research is the ability to derive the possible product compositions and the column structure for not only a fixed feed composition but also for the entire feed region. For each example process, the entire feed composition region is divided into several subregions, each of which has similar characteristics for the product composition and the column structure. The list of column configurations and product specifications for the entire feed region can be used effectively for the screening of undesirable feed compositions and column configurations. The conversion rate and the distillation boundary play an important role for the achievable product compositions. In this study, mainly the case of maximal possible conversion is discussed. Nonetheless, it is possible to calculate the feasibility of the process for a lower extent of conversion. Results for a lower extent of conversion are explained for the productions of acetic acid and ethyl acetate. The drawback of the method under study is that it assumes both an infinite separation efficiency and a fixation of one of the product compositions. The assumption of infinite flow rate inside the column makes the calculation of the operating line much easier. Nonetheless, this assumption may eliminate the possibility that the operating line with reaction crosses the conventional distillation boundary. The assumption of the fixation of one of the product compositions simplifies the determination of the feasibility of the given degree of conversion. Nonetheless, the possibility that a feasible operating line exists for other product compositions cannot be disregarded. Therefore, the final decision of the feasibility of the system should be made based upon more precise simulations or experimental studies. It should be noted that even though SA cannot guarantee whether the derived system is really feasible or not, it provides valuable hints for the first step of the design of the process. The recent existing SA design program for RD, developed by Danilov,18 is in need of much revision, such as recognition of multiple products, multiple feeds, and multicomponent reactions with various azeotropes. So far, SA has only been applied to direct and indirect separation. At present, the theory of SA is being further developed, and the above-stated revisions will be considered. The most significant advantage of the proposed method over other design tools is that the product compositions, the extent of the reaction, and the number of stages need not be fixed a priori. Moreover, it is even possible to obtain useful information regarding the length and location of the reactive zone. The results of various examples indicate that SA is a very convenient tool that provides answers to the question of feasibility and hints for the early stage of design. Nomenclature AA ) acetic acid AcAn ) acetic anhydrid B ) butanol B1 ) bottom product BA ) butyl acetate bp ) boiling point C ) cumene

4066

Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999

D/B ) amount of distillate/amount of bottom product D1 ) distillate E ) ethanol EA ) ethyl acetate EG ) ethylene glycol EO ) ethylene oxide LCI ) line of chemical interaction LMB ) line of mass balance CEM ) chemical equilibrium manifold LSS ) limiting steady state M ) methanol MA ) methyl acetate MTBE ) methyl-tert-butyl ether TAME ) tert-amyl-methyl-ether P ) propylene RD ) reactive distillation SA ) static analysis W ) water X ) liquid composition X*, X** ) pseudoinitial mixture () steady state)

c. Production of Ethylene Glycol. The equilibrium constant has the value 1.108. It is determined from thermodynamic data, taken from the Handbook of Chemistry 5-4, 10th ed. Antoine equation coefficients and NRTL coefficients are shown in Tables 9 and 10.

Indices

Table 10. NRTL Coefficients for the Production of Ethylene Glycola

Table 8. NRTL Coefficients for the Production of Cumene at 20 bara ∆G12/R ∆G21/R R12 a

component

A

B

C

ethylene oxide water ethylene glycol

16.74 18.30 20.25

2567.61 3816.44 6022.18

-29.01 -46.13 -28.25

a

a

A

B

C

acetic anhydride water acetic acid

16.40 18.30 16.81

3287.56 3816.44 3405.57

-52.36 -46.13 -56.34

a

From Reid and

EO-W

W-EG

EO-EG

793.90 230.48 0.30

1888.75 -1445.47 0.08

279.28 -34.72 0.8

From ProII.

Table 11. Antoine Equation Coefficients for the Production of Ethyl Acetatea

a

component

A

B

C

ethanol acetic acid ethyl acetate water

18.91 16.81 16.15 18.30

3804.0 3405.6 2790.5 3816.4

-41.68 -56.34 -57.15 -46.13

From Reid and Prausnitz.25

Prausnitz.25

Table 6. NRTL Coefficients for the Production of Acetic Acida ∆G12/R ∆G21/R R12

From Reid and Prausnitz.25

d. Production of Ethyl Acetate. The Antoine equation coefficients and NRTL coefficients are shown in Tables 11 and 12. Conditions of the azeotropes are shown in Table 13.

Table 5. Antoine Equation Coefficients for the Production of Acetic Acida component

C-P -26.116 54 -6.002 99 -0.979 32

From ProII.

∆G12/R ∆G21/R R12

Physical data and parameters used in the calculations are summarized in this section. a. Production of Acetic Acid. The equilibrium constant has the value 1. It is determined from thermodynamic data, taken from the Handbook of Chemistry 5-4, 10th ed. Antoine equation coefficients and NRTL coefficients are shown in Tables 5 and 6.

B-C 964.0 -407.0 0.3693

Table 9. Antoine Equation Coefficients for the Production of Ethylene Glycol

* ) pseudoinitial F ) feed

Appendix

B-P 38.671 71.0346 0.4628

AcAn-W

AA-AcAn

AA-W

319.00 252.91 -0.91

21.64 282.84 0.30

-400.56 622.39 0.1

Results of Kogan27 and ProII were used for regression of the UNIFAC values into the NRTL equation.

Table 12. NRTL Coefficients for the Production of Ethyl Acetatea E-AA

E-EA

E-W

AA-EA

AA-W

EA-W

∆G21/R 254.63 158.35 315.17 97.48 -427.90 523.16 ∆G21/R -457.75 155.91 7.13 -103.57 400.60 1059.54 R12 0.25 0.25 -0.67 0.30 0.10 0.37 a

From Danilov.18

a

Table 13. Conditions of the Azeotropes for the Production of Ethyl Acetatea

b. Production of Cumene. The Antoine equation coefficients and NRTL coefficients are shown in Tables 7 and 8. Table 7. Antoine Equation Coefficients for the Production of Cumenea

a

a

component

A

B

C

propylene benzene cumene

15.70 15.90 15.97

1807.53 2788.51 3363.60

-26.15 -52.36 -63.37

From Reid and Prausnitz.25

E-EA-W E-EA E-W EA-W

composition [mol %]

bp [°C]

9.5/61.8/28.7 45.0/55.0 91.7/8.3 69.0/31.0

70.3 71.8 78.0 70.4

From Ogorodnikov et al.26

e. Production of Butyl Acetate. The Antoine equation coefficients and NRTL coefficients are shown in Tables 14 and 15. Conditions of the azeotropes are shown in Table 16.

Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999 4067 Table 14. Antoine Equation Coefficients for the Production of Butyl Acetatea butanol acetic acid butyl acetate water a

A

B

C

10.34 11.35 9.84 11.73

3009.6 4161.9 3323.4 3852.2

-99.52 -24.27 -61.03 -44.36

From Gmehling and Onken.28

Table 15. NRTL Coefficients for the Production of Butyl Acetatea W-B

W-AA

W-BA

B-AA

B-BA AA-BA

∆G12/R 1249.9 -400.56 2369.69 354.03 19.76 910.82 ∆G21/R 220.19 622.39 -429.27 -474.91 164.02 -706.24 R12 0.425 0.1 0.1004 0.2 0.45 0.1 a

From Danilov.18

Table 16. Conditions of the Azeotropes for the Production of Butyl Acetate at 32 kPaa W-B W-BA B-BA B-AA W-B-BA B-AA-BA a

Comp. [mol %]

bp [°C]

79.1/20.9 72.0/28.0 56.6/43.4 63.5/36.5 68.8/5.9/25.3 22.2/26.3/51.5

66.6 63.8 87.1 91.8 63.5 88.3

From Danilov.18

Literature Cited (1) Agreda, V. H.; Partin, L. R.; Heise, W. H. High-purity methyl acetate via reactive distillation. Chem. Eng. Prog. 1990, 86, 40. (2) Saito, S.; Michishita, T.; Maeda, S. Separation of meta- and para-xylene mixture by distillation accompanied by chemical reactions. J. Chem. Eng. Jpn. 1971, 4, 37. (3) DeGarmo, J. L.; Parulekar, V. N.; Pinjara, V. Consider reactive distillation. Chem. Eng. Prog. 1992, 88, 43. (4) Ignatius, J.; Jaervelin, H.; Lindquist, P. Use TAME and heavier ethers to improve gasoline properties. Hydrocarbon Process. 1995, 2, 51. (5) Ruiz, C. A.; Basualdo, M. S.; Scenna, N. J. Reactive distillation dynamic simulation. Trans. Inst. Chem. Eng. Part A 1995, 73, 363. (6) Fidkowski, Z. T.; Malone, M. F.; Doherty, M. F. Feasibility of separations for distillation of nonideal ternary mixtures. AIChE J. 1993, 39, 1303. (7) Barbosa, D. A. Distillation of Reactive Mixtures. Ph.D. Dissertation, University of Massachusetts, Amherst, MA, 1987. (8) Buzad, G. Design of Kinetically Controlled Reactive Distillation Columns. Ph.D. Dissertation, University of Massachusetts, Amherst, MA, 1994. (9) Doherty, M. F.; Buzad, G. Reactive distillation by design. Chem. Eng. Res. Des. 1992, 70, 448.

(10) Bessling, B.; Schembecker, G.; Simmrock, K. H. Design of Processes with Reactive Distillation Line Diagrams. Ind. Eng. Chem. Res. 1997, 36, 3032. (11) Nisoli, A.; Malone, M. F.; Doherty, M. F. Attainable Regions for Reaction with Separation. AIChE J. 1997, 43, 374. (12) Serafimov, L. A.; Zharov, V. T.; Timofeyev, V. S. Rectification of multicomponent mixtures. I. Topological analysis of liquidvapor phase equilibrium diagrams. Acta Chim. Acad. Sci. Hung. 1971, Tomus 69, 4, 383. (13) Serafimov, L. A. In Reactive-Mass Exchange Processes, Problems and Prospects. Proceedings of Chemical Engineering Science for Advanced Technologies, CESAT2, Moscow, 1996. (14) Pisarenko, Y. A.; Serafimov, L. A. Steady states for a reactive-distillation column with one product stream. Theor. Found. Chem. Eng. 1988, 21, 281. (15) Pisarenko, Y. A.; Danilov, R. Y.; Serafimov, L. A. InfiniteEfficiency Operating Conditions in Analysis of the Statics of Reactive Rectification. Theor. Found. Chem. Eng. 1995, 29, 556. (16) Pisarenko, Y. A.; Danilov, R. Y.; Serafimov, L. A. Possible modes of separation in continuous reactive-rectification processes. Theor. Found. Chem. Eng. 1996, 30, 585. (17) Pisarenko, Y. A.; Danilov, R. Y.; Serafimov, L. A. Application of the concept of limiting paths to estimating the feasibility of steady states in analyzing the statics of reactive-rectification processes. Theor. Found. Chem. Eng. 1997, 31, 43. (18) Danilov, R. Y. Development of an Automatic Variant of Analysis of the Statics of Continuous Reactive Distillation processes. Ph.D. Dissertation, Lomonosov Moscow State Academy of Fine Chemical Technology, Moscow, 1997 (in Russian). (19) Giessler, S.; Pisarenko, Y. A.; Danilov, R. Y.; Serafimov, L. A.; Hasebe, S.; Hashimoto, I. Feasibility study of reactive distillation using the analysis of the statics. Ind. Eng. Chem. Res. 1998, 37, 4375. (20) Shoemaker, J. D.; Jones, E. M. Cumene by catalyst distillation. Hydrocarbon Process. 1987, 57-58. (21) Corrigan, T. E.; Miller, J. H. Effect of distillation on a chemical reaction. Ind. Eng. Chem. Process Des. Dev. 1968, 7, 3. (22) Parker, A. S. U. S. Patent 2,839,588, 1958. (23) Ciric, A. R.; Miao, P. Steady-State Multiplicities in an Ethylene Glycol Reactive Distillation Column. Ind. Eng. Chem. Res. 1994, 33, 2738. (24) Dudukovic, M. P. Steady-state multiplicities in an ethylene glycol reactive distillation column. Ph.D. Dissertation, University of Cincinnati, Cincinnati, OH, 1994. (25) Reid, R.; Prausnitz, J. M. The Properties of Liquids and Gases, 4th edition, McGraw-Hill, New York, 1987. (26) Ogorodnikov, S. K.; Lesteva, T. M.; Kogan, V. B. Azeotropic Mixtures Directory; Khimiya: Leningrad, 1971; 848 (in Russian). (27) Kogan, V. B.; Fridman, V. M.; Kafarov, V. V. Equilibrium between Liquid and Vapor; Nauka: Moscow-Leningrad, 1966. (28) Gmehling, J.; Onken, U. Vapor-Liquid Equilibrium Data Collection, Chemistry Data Series; DECHEMA: Frankfurt, 1977.

Received for review January 4, 1999 Revised manuscript received July 8, 1999 Accepted July 20, 1999 IE9900162