Você está na página 1de 10

Chemical Engineering Science 57 (2002) 1511 1520

www.elsevier.com/locate/ces

Synthesis of cylohexanol by three-phase reactive distillation: in uence of kinetics on phase equilibria


Frank Steyera , Zhiwen Qia , Kai Sundmachera; b; ,
a Max-Planck-Institute b Process

for Dynamics of Complex Technical Systems, Sandtorstra e 1, D-39106 Magdeburg, Germany Systems Engineering, Institut fur Verfahrenstechnik, Otto-von-Guericke-University Magdeburg, Universitatsplatz 2, D-39106 Magdeburg, Germany

Abstract A reactive distillation process is being suggested for the production of the commercially interesting intermediate cyclohexanol from cyclohexene and water, which avoids some of the drawbacks of the conventional liquid-phase cyclohexane oxidation process, especially with respect to operational safety. This reactive distillation process has its own intricate challenges due to the fact that reaction, distillative separation and liquid-phase splitting occur simultaneously. The interaction of these three phenomena is studied using residue curve maps for both, model simulations and experimental data. Based on the xed point analysis of the residue curves, a novel process is proposed for the reactive separation of cyclohexene=cyclohexane mixtures which is di cult to be carried out by conventional distillation due to very close boiling temperatures. ? 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Reactive distillation; Residue curve maps; Multiphase reactors; Phase equilibria; Reactive separation

1. Introduction Cyclohexanol is being synthesized on an industrial scale today as an intermediate in the production of adipic acid and -caprolactame. These in turn are intermediates for the production of Nylon 6,6 and Nylon 6. Since Nylon is a bulk polymer the world production of cyclohexanol had already reached 1.1 million tons per year in the beginning of the 1980s (Industrial Organic Chemicals, 1999). Currently there are three commercial routes to produce cyclohexanol. The route to produce over 90% of western European and the US cyclohexanol is based on the hydrogenation of benzene to cyclohexane and its subsequent oxidation by air to cyclohexanol=cyclohexanone. This process su ers from several disadvantages. Besides the high consumption of hydrogen for the hydrogenation step and the associated high energy demands of the process, the oxidizing step has the disadvantage of producing multiple by-products. The main disadvantage of the oxidizer is the risk with respect to its operational safety. Since cyclohexane has to be
Corresponding author. Process Systems Engineering, Institut fur Verfahrenstechnik, Otto-von-Guericke-University Magdeburg, Universitatsplatz 2, D-39106 Magdeburg, Germany. Tel.: +49-391-671-8704; fax: +49-391-6711245. E-mail address: kai.sundmacher@vst.uni-magdeburg.de (K. Sundmacher).

mixed with air, there is a zone with an explosive mixture within the reactor. This has resulted in an explosion of an oxidizer in Flixborough, England in 1974 with several fatalities and the complete destruction of the plant (Industrial Organic Chemicals, 1999). An alternative still being used to a smaller extent in western Europe and the US is the hydrogenation of phenol. The attractiveness of this process is directly coupled to the availability and price of phenol in comparison to benzene. Also this process su ers from the high hydrogen consumption for the hydrogenation step. The third process currently in use by Asahi Chemical Co. also starts with benzene which is hydrogenated to cyclohexene. This saves one-third of the input hydrogen normally necessary for the conversion to cyclohexane. Cyclohexene is then hydrated to cyclohexanol using a strongly hydrophilic zeolite catalyst (H-ZSM 5, high Si=Al ratio) which avoids the formation of signicant amounts of by-products (Mitsui, Osamu, Fukuoka, & Yohei, 1986). This process also overcomes the operational safety problems associated when mixing air with the organic reactant by entering the necessary hydroxyl group in the form of water. Asahi is currently operating plants in the 10,000 100,000 ton=yr range using this technology. The process consists of a slurry reactor with a subsequent decanter and a distillation column. Water and cyclohexene are fed to the reactor together with

0009-2509/02/$ - see front matter ? 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 9 - 2 5 0 9 ( 0 2 ) 0 0 0 2 3 - 4

1512

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520

Cyclohexane Decanter Water

OH

+ H 2O

Catalytic Section

Cyclohexene/Cyclohexane

Cyclohexanol
Fig. 1. Flowsheet of the proposed reactive distillation column.

recent papers dealing with this subject were written on the butyl acetate esterication system and only mentioned phase-splitting behaviour as a side phenomenon (Venimadhavan, Malone, & Doherty, 1999; Zhicai, Xianbao, & Jing, 1998; Loning, Horst, & Ho man, 2000). Westerberg, Lee, and Hauan (2000) gave a good survey over the possible combinations of many phenomena including the three involved in this case, but they did not deal in detail distillation with the occurrence of liquid-phase splitting in reactive distillation. Gumus and Ciric (1997) specically tried to simulate such a system. Their focus was on the simultaneous reaction=distillation=phase-splitting behaviour because its prediction is computation time intensive so that it is ideally suited to show the e ciency of a new numerical algorithm. Recently, Qi, Kolah, and Sundmacher (2002) have started to systematically investigate reactive distillation with liquid-phase splitting considering model reaction systems as well as real reaction systems. In this paper, the in uence of the reaction kinetics and the phase equilibria on the reactive distillation process has been studied by both, model simulations and experimental work. Based on these investigations, novel processes are proposed to produce cyclohexanol by hydration of cyclohexene, or to separate cyclohexene=cyclohexane mixtures by reactive distillation. 2. Residue curve maps of the nonreactive mixture As a rst step towards the design of a reactive distillation process, the nonreactive vapourliquid equilibria were studied by means of residue curve maps for the ternary system cyclohexene=water=cyclohexanol. 2.1. Experimental A residue curve describes what happens to the composition of the liquid residue in an open batch vessel as more and more liquid is evaporated. From any starting point within the composition space of a given system, the concentrations will change towards the highest boiler as more and more lighter boiling components are evaporated. Consequently, the boiling temperature in the batch still will increase. A residue curve map is a compilation of such residue curves with different starting points (usually near the lightest boiler). Since such a system can be easily simulated and is also easy to set up experimentally, residue curve maps are very often used as a tool not only to predict the potential products of reactive and nonreactive distillation columns but in addition, they can also be directly used to experimentally validate physicochemical models for reactive distillation processes. The experimental set-up used in this work (Fig. 2) is a 0:5 l glass batch vessel which is stirred at 600 rpm with a glass propeller stirrer and heated by an oil-lled heating mantle to the boiling point. The stirring speed was adjusted to assure good mixing and very ne droplets during the

very ne catalyst particles ( 1 m). The resulting slurry is separated in the decanter leading to an organic phase and an aqueous phase. The catalyst particles are recovered quantitatively in the aqueous phase which is recycled back to the slurry reactor. The organic phase is composed of cyclohexanol as the reaction product, unreacted cyclohexene as well as benzene and cyclohexane which are impurities coming from the preceding benzene hydrogenation step. Due to the very close boiling points of benzene (80:10 C), cyclohexene (82:98 C) and cyclohexane (80:74 C), they are di cult to separate by conventional distillation (Industrial Organic Chemicals, 1999). Cyclohexanol, on the other hand, has a signicantly higher boiling point of 161:1 C at normal pressure and is therefore easily separated from the rest of the mixture in a distillation column. This rest is then partly recycled back to the slurry reactor. Taking into account that the cyclohexene hydration is slightly exothermic and is equilibrium limited it is presumably well suited to be carried out in a reactive distillation column. This reactive distillation column would replace the slurry reactor, the decanter and the distillation column in the conventional process. The proposed novel process is illustrated in Fig. 1. From a more fundamental perspective, this system is also of major interest due to the fact that cyclohexene and water have a very limited mutual miscibility. Only little research has been done on reacting systems with simultaneous distillation and liquid-phase splitting. Most

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520

1513

M GC GC

' HD

'

'' HD

H ''

used see Qi et al. (2002). For the explanation of the symbols used in Fig. 3 please refer to Table 1. As can be seen, the experimental data do not agree very well with the simulated residue curves. Also, there are signicant deviations between the simulated azeotropic points and the corresponding literature data. To understand why there are such substantial deviations, one has to recall that the system investigated here is known to undergo phase splitting. This was also observed in the experiments, but was not considered in the simulated residue curves shown in Fig. 3. 2.3. Liquidliquid equilibria (LLE)

Fig. 2. Experimental setup for the measurements of phase equilibria with and without chemical reaction.

measurements in order to avoid any possible mass transfer limitation. The vapour is condensed by a distillation bridge mounted on the vessel and is collected in a decanter that allows to measure the volumes of the two liquid phases. Alternatively to the distillation bridge, a total re ux cooler can be attached to the top of the vessel. The distillation bridge and re ux cooler are necessary to prevent the resulting vapour from forming explosive mixtures in the exhaust system. Since liquid-phase splitting was expected, a method had to be developed to determine the molar holdups of the two phases, H and H , within the batch vessel. Only with the knowledge of these holdups and their compositions, the average liquid-phase composition can be calculated. For the determination of the holdups, a mass-balance based approach was applied. For this purpose, the initial amounts of all three components were noted. As the batch experiment went on, the holdups and the compositions of the two phases in the decanter were measured. This allowed to calculate the amounts of the substances left within the vessel by means of mass balances. The compositions of the two liquid phases within the batch vessel were directly measured. Analysis of the samples drawn from the batch and the decanter was done with a gas chromatograph (GC, HP 6890, HP Innowax column 30 m; 250 m diameter) equipped with a TCD and an FID. On account of the high di erence in boiling points, a GC temperature program was applied start ing at 50 C for 3 min and then raising the temperature to 170 C at a heating rate of 30 C=min. This temperature was then held for 3 min to clean the GC column. The chemicals used were cyclohexene ( 99%) and cyclohexanol ( 99%), both from Merck, and deionized water. 2.2. Vapourliquid equilibria (VLE) First, residue curves were simulated using the VLE model published by Doherty and Perkins (1978). The UNIFAC group contribution method was used to predict the liquid phase activity coe cients. For the interaction parameters

To take this e ect into account, additional measurements and more detailed simulations were carried out. Again the UNIFAC method was used to estimate the activity coe cients. The predicted binodal curves and liquidliquid tie lines are shown in Fig. 4. For comparison, the experimental data as well as a literature data point representing the solubility of water in cyclohexanol were inserted into the diagram. The agreement between UNIFAC-based predictions and the experimental data is satisfactory on a qualitative basis. On the quantitative side, however, some discrepancies have to be noticed. The predicted solubility of water in pure cyclohexene (xWater =0:019) deviates signicantly from the experimentally observed solubility (xWater 0:005). On the other hand, the solubility of water in pure cyclohexanol is higher than predicted. These deviations between UNIFAC-based calculations and experimental results are signicant in their size, which is due to the fact that the UNIFAC group contribution model does not rely on any data measured directly at the special substances considered here. However, the model gives correct qualitative results, which should be a reasonable basis for the prediction of phase splitting under conditions of reactive distillation, too. The scattering of the experimental data towards the cyclohexanol corner in Fig. 4 is due to a tendency of water=cyclohexanol mixtures to form very small and relatively stable droplets due to their small di erence in densities ( ol = 0:94 kg=l; H2 O = 1:00 kg=l). To reach phase equilibrium quickly, the mixtures were agitated strongly. Since even a small droplet of water in the cyclohexanol phase can signicantly change the measured water solubility, some scattering of the experimental data always occurs. 2.4. Vapourliquidliquid equilibria (VLLE) From the results reported in Section 2.3, one can conclude that the system cyclohexene=water=cyclohexanol shows a strongly nonideal mixing behaviour in the liquid phase which leads to an extremely broad miscibility gap. Therefore, a VLLE calculation has to be implemented into the residue curve simulations. This is outlined in the following. The component mass balances for the considered

1514

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520


Water 100 C 1 97.8 C 0.9 Azeotrope 1 96.42 C 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.1 Cyclohexanol 161.1 C 70.8 C Azeotrope 2 59.75 C Distillation Boundary

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9 1 Cyclohexene 82.9 C

Fig. 3. Residue curve map for nonreactive vapourliquid equilibrium system with the experimental data and the literature azeotropes (p = 105 Pa). Table 1 Legends of symbols and curves for all ternary diagrams in this work

Legend of symbols

Meaning of symbols Stable node Unstable node Saddle point Literature dataa Experimental data (this work)

Legend of curves -- ------

Meaning of curves Residue curve Distillation boundary Binodal curve Chemical equilibrium curve

4
? Open symbols
a Gmehling,

Menke, Kranfczyk, and Fisher (1994); Industrial Organic Chemicals (1999).

batch distillation are given by d xi = xi yi d (1)

with xi as the average liquid-phase mole fraction and yi as the vapour-phase mole fraction, d = (V=H ) dt as the dimensionless time, V as the vapour ow rate, and H as the total molar liquid holdup in the still. Eq. (1) is subject to the initial conditions xi ( = 0) = xi; 0 : (2)

compositions expressed in terms of the vectors of mole fractions, x and x , in the respective phase and on the temperature T: P and pisat are the total pressure of the system and the saturation pressure of component i at the boiling temperature, respectively. The equality between the two right parts of Eq. (3) is based on the phase equilibrium condition between the two liquid phases, i.e. xi i (x ; T ) = xi
i

(x ; T ):

(4)

The ratio of the molar organic phase holdup (H ) to the total molar liquid holdup (H + H ) is dened as = H : H +H (5)

The mole fractions in the vapour phase have to satisfy the relations xi i (x ; T )pisat (T ) xi i (x ; T )pisat (T ) = : (3) P P In Eq. (3), the activity coe cients of the two liquid phases, i and i are dependent on the corresponding liquid-phase yi =

Therefore, the average liquid mixture mole fractions are given by xi = xi + (1 )xi : (6)

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520

1515

Water 100 C 1 0.9 0.8 0.7


x 10
_3

0.6 0.5 0.4 0.3 0.2 0.1 Binodal Curve at T=450 K L-L Tie Lines

Binodal Curve at T=298 K 0 0 0.2 0.4 Cyclohexanol 161.1 C

0.6

0.8

1 Cyclohexene 82.9 C

Fig. 4. Phase diagram for the liquidliquid equilibrium at isothermal conditions.

The mole fractions in all three phases have to obey the following summation equations:
NC NC NC

cancel each other out as can be seen from the data in Table 2. As a consequence, the chemical equilibrium constant Ka (298:15 K) = exp
0 R G298:15; l = 5:6 R 298:15

yi = 1;
i=1 i=1

xi = 1;
i=1

xi = 1:

(7ac)

From this set of equations a new set of residue curves were simulated as shown in Fig. 5, where the experimental data points and the reference data from the literature indicating the azeotropic points are plotted, too. As can be seen, the simulated liquid-phase compositions agree much better with the measured data points than those which were predicted by VLE only. But there still exists a discrepancy between the predicted azeotropes and the corresponding literature values. These di erences can presumably be attributed to the use of UNIFAC as the activity coe cient model with its drawbacks as mentioned before. 3. Residue curve maps of the reactive mixture 3.1. Chemical equilibrium As stated before, the reaction of cyclohexene with water to cyclohexanol is signicantly limited by its chemical equilibrium. When trying to predict the reaction equilibrium, some di culties are encountered because the Gibbs enthalpies of formation for the reaction components nearly

(8)

is di cult to x exactly due to the uncertainty of the standard Gibbs enthalpy of reaction R G 0 . However, the measurements of Panneman and Beenackers (1992) seem to conrm the Ka value in Eq. (8). The temperature dependence of the equilibrium constant was estimated by assuming that the reaction enthalpy R H 0 is independent on temperature in the range of boiling temperatures which can be expected at ambient pressure. This allows the use of the following equation given by Panneman and Beenackers (1992): Ka (T ) = 2:37 105 exp 30;236 : RT (9)

As Qi et al. (2002) have shown, phase splitting has a strong in uence on the chemical equilibrium curve. As can be seen in Fig. 6i, the equilibrium curve follows a unique line within the region of liquidliquid phase splitting. This is because only mixtures located on this unique tie line fulll both the chemical equilibrium condition and the VLLE conditions.

1516

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520


Water 100 C
1

97.8 C Azeotrope 1 96.21 C


0.9

0.8

Distillation Boundary

0.7

0.6

0.5

0.4

Azeotrope 2 58.84 C 70.8 C

0.3

0.2

0.1

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Cyclohexanol 161.1 C

Cyclohexene 82.9 C

Fig. 5. Residue curve map for the nonreactive vapourliquidliquid system with the experimental data and the literature azeotropes (p = 105 Pa). Table 2 Thermodynamic data for the system cyclohexene=water=cyclohexanol
0 a f H298:15; l ; kJ=mol 0 f S298:15; l ; kJ=mol

Ka

0 f G298:15; l ; kJ=mol

Cyclohexene Water Cyclohexanol

37:82 285:83 352:00


0 R H298:15; l

214.60 69.95 203.87 = 28:35 kJ=mol

101:803 306:686 412:784


0 R G298:15; l

= 4:295 kJ=mol

a Lindstrom

and Mallard (2001).


phase = kf (T ) aene aH2 O

3.2. Reaction kinetics To allow for the e ect of the reaction kinetics on the reactive distillation process with phase splitting, the batch still model is modied taking into account that a reaction can occur in both the liquid phases. In general, the chemical reaction can be expressed by
cr

1 aol Ka

(11)

| r |Ar
r=1

kb p=1

kf

cp

| p |Ap ;

(10)

where Ai represents the reacting species, r and p are indices for the reactants and products, and i are the stoichiometric coe cients ( i 0 for educts, i 0 for products). The overall sum of stoichiometric coe cients is T ; cr and cp are the total number of reactants and products, respectively. The reaction rate in each liquid phase is expressed as cp cr 1 | | phase | r phase = kf (T ) ar r | ap p Ka
r=1 p=1

phase with kf as the temperature-dependent rate constant of the forward reaction in the considered liquid phase. Due to the fact that the liquid phase activities ai in the two liquid phases are equal at liquidliquid equilibrium conditions, in Eq. (11), the reaction rates in the two liquid phases can di er phase in case of di erent rate constants kf , only. These rate constants can be di erent due to di erent catalyst activities in the two liquid phases, e.g. by solvent e ects, or=and due to di erent catalyst holdups as a result of the fact that the catalyst itself is distributed between the two liquid phases. phase The temperature dependency of kf is described by the Arrhenius equation phase phase kf = kf; ref exp phase EA R

1 1 T Tref

(12)

where EA stands for the activation energy in the considered liquid phase.

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520


Water 1 0.9 0.8 0.7 Water 1 0.9 0.8 Water 1 0.9 0.8

1517

(g)

0.7 0.6 0.5 0.4 0.3 0.2 0.1

(h)

0.7 0.6 0.5 0.4 0.3 0.2 0.1

(i)

500 (fast)

0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 Cyclohexanol Water 1 0 0.3 0.4 0.5 0.6 0.7 0.8

0 0.9 1 0 0.1 0.2 Cyclohexene Cyclohexanol Water 1 0.9 0.8

0.3

0.4

0.5

0.6

0.7

0.8

0 0 0.1 0.2 0.9 1 Cyclohexene Cyclohexanol Water 1 0.9 0.8

0.3

0.4

0.5

0.6

0.7

0.8

0.9 1 Cyclohexene

Da (reaction rate watery phase)

0.9 0.8

0.05 (moderate)

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 Cyclohexanol Water 1 0.9 0.8 0.7 0 0.3 0.4 0.5

(d)

0.7 0.6 0.5 0.4 0.3 0.2 0.1

(e)

0.7 0.6 0.5 0.4 0.3 0.2 0.1

(f)

0.6

0.7

0.8

0.9 1 0 0.1 0.2 Cyclohexene Cyclohexanol Water 1 0.9 0.8

0.3

0.4

0.5

0.6

0.7

0.8

0 0.9 1 0 0.1 0.2 Cyclohexene Cyclohexanol Water 1 0.9 0.8

0.3

0.4

0.5

0.6

0.7

0.8

0.9 1 Cyclohexene

(a)

0.7 0.6 0.5 0.4 0.3 0.2 0.1

(b)

0.7 0.6 0.5 0.4 0.3 0.2 0.1

(c)

0 (none)

0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 Cyclohexanol 0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Cyclohexene

0 0.1 0.2 Cyclohexanol

0.3

0.4

0.5

0.6

0.7

0.8

0 0.9 1 0 0.1 0.2 Cyclohexene Cyclohexanol

0.3

0.4

0.5

0.6

0.7

0.8

0.9 1 Cyclohexene

0 (none)

0.05 (moderate) Da (reaction rate organic phase)

500 (fast)

Fig. 6. Residue curve maps at di erent Damkohler numbers in the two liquid phases (p = 105 Pa).

3.3. Mass balances If the occurrence of the chemical reaction in the two liquid phases is incorporated into the component mass balances, they take the following form: d xi H = (xi yi ) + ( i d V d xi H = (xi yi ) + ( i d V
T xi )[r

By dening the phase Damkohler number Daphase


phase H0 kf; ref

V0

(16)

the mass balances, Eq. (14), can be rewritten as d xi H V0 = (xi yi ) + ( i d V H0 Da kf kf; ref
T xi )

+ (1 )r ]:

(13)

Combining Eq. (13) with the rate expression Eq. (11) yields
T xi )[kf

+ (1 )kf ]R (14)

+ (1 )Da

kf kf; ref

R:

(17)

with R as the dimensionless reaction rate: aol R = aene aH2 O : Ka

(15)

In this work, the heating policy V=V0 = H=H0 was applied that holds the basic qualities of residue curves also for other heating policies (Venimadhavan et al., 1994) and can also be

1518

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520

realized experimentally. Then, the component mass balances can be given in the following simplied form: d xi = (xi yi ) + ( i T xi ) d Da kf kf; ref + (1 )Da kf kf; ref R: (18)

For the determination of the ratios of the rate constants, phase phase kf =kf; ref , Eq. (12) was used. The parameter values were adopted from Panneman and Beenackers (1992). The activation energies were assumed to be equal for the two phases, i.e. EA and EA . 3.4. Discussion of simulated results Using Eq. (18), a systematic simulation study was performed at di erent Damkohler numbers in the organic and aqueous phase, Da and Da , respectively. These Damkohler numbers were set to 0 (no reaction), 0.05 (moderately fast reaction), and 500 (very fast reaction approaching chemical equilibrium). The resulting nine residue curve maps are shown in Fig. 6. For the symbols and curves used in these diagrams please refer to Table 1. To explain the di erent e ects involved, rst the three diagrams with equal Da numbers in the two liquid phases, i.e. Da = Da , are discussed (Figs. 6a, e, and i). When the reaction rates in both phases are zero, the nonreactive distillation behaviour is recovered as shown in Fig. 6a. In this case the possible stable nodes are cyclohexanol and water, the unstable node is the azeotrope between water and cyclohexene (Azeotrope 2). Moreover, pure cyclohexene and the azeotrope between cyclohexanol and water (Azeotrope 1) are saddle points. When a moderately fast reaction occurs in both phases, as shown in Fig. 6e, cyclohexanol is no longer a stable node because it is decomposed into cyclohexene and water. Therefore, the stable node moves out of the cyclohexanol corner. Due to the appearance of a second distillation boundary in the cyclohexene corner with the onset of reaction, now pure cyclohexene is an additional stable node. At very high Da numbers in both phases (Fig. 6i), the upper distillation boundary disappears. This is due to the fact that at a critical Damkohler number, i.e. Da = Da = 0:109 (see Qi et al., 2002), the saddle point which is responsible for the existence of the distillation boundary coincides with the stable node coming from the cyclohexanol corner and they cancel each other. Moreover, pure water and pure cyclohexene remain stable nodes. As can be seen quite well in this diagram the chemical equilibrium curve is a straight line within the two-phase region. Now, the cases (see Figs. 6b, c, d, and g) in which the chemical reaction takes place only in one of the two liquid phases are discussed. One should be aware that these cases are only theoretical extremes. In practice, very often an applied catalyst is not only present or active in one

single phase, but it has a certain activity in each phase. However, these theoretical extremes may help to understand the principal behaviour of reactive distillation at conditions of liquid-phase splitting. In the four involved diagrams, the reaction only takes place as long as the reactive phase exists. Outside the two-phase region in the nonreactive phase, the pure distillation behaviour is recovered. Inside the two-phase region, the residue curves are qualitatively similar to those that are obtained if the reaction takes place in both liquid phases at the same Da numbers. But their quantitative properties such as the trajectories, locations of the singular points and separatrices, are di erent. Moreover, note that the aqueous phase region close to the water corner is almost invisible in the ternary diagrams whereas the organic phase region is quite large. Therefore, compared to the case that the reaction occurs in both liquid phases at the same Da, the residue curve maps are qualitatively and quantitatively similar to the diagram where the reaction occurs only in the organic phase (compare Figs. 6b and c to Figs. 6e and i), but they are quite di erent when the chemical reaction occurs only in the aqueous phase (compare Figs. 6d and g to Figs. 6e and i). At very high reaction rates (Da = 500, see Fig. 6g), there is a distillation boundary partly coinciding with the binodal curve at the organic side, which intersects with the chemical equilibrium curve at a certain point. Below this distillation boundary, pure cyclohexanol is the stable node, the boundary pure water above is the stable node. Of the two gures which show reaction with di erent Da numbers in the two liquid phases (Figs. 6f and h), the one with a high reaction rate in the organic phase (Fig. 6f) is almost indistinguishable from the case of high reaction rates in both phases (Fig. 6i). This is because the aqueous phase region is so small that the system behaviour for reactive distillation behaviour di ers very little from that for the nonreactive conditions. Fig. 6h shows an interesting type of behaviour. In the organic phase region below the binodal curve, the same behaviour as in the case with equal moderate Da numbers (Fig. 6e) is retained. In the two-phase region due to the very high Da number in the aqueous phase, Da = 500, all trajectories move straight towards the chemical equilibrium line. For those trajectories which meet the chemical equilibrium line within the two-phase region, the nal product will be pure water as the stable node. For those trajectories which cross the binodal curve before reaching the chemical equilibrium curve, the cyclohexanol-rich stable node denes the nal product composition. Between the attractive domains of these two stable nodes, there is a separatrix. 4. Conceptual design of continuous processes The residue curve maps discussed in the previous section can be used for the conceptual design of a continuous reactive distillation process for the production of cyclohexanol. Due to the existence of a stable node near the cyclohexanol

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520

1519

Cyclohexane Water

Catalytic Section Forward Reaction (Da < 0.109) Cyclohexene/ Cyclohexane Catalytic Section Back Reaction (Da > 0.109) Cyclohexanol Cyclohexene

Fig. 7. Coupled process to the reactive separation of cyclohexene=cyclohexane mixtures.

corner at low Da numbers, the operation at these reaction rates in the organic phase is advisable because this allows to overcome chemical equilibrium limitations. Especially, one should be careful not to achieve Da numbers above the critical value 0.109. Since pure cyclohexanol is not a stable node in any case if the chemical reaction takes place in the organic phase, a nonreactive stripping section in the column will be required. Evidently, a hybrid column is necessary to achieve the process goals. At the top of the column nonreacted water and cyclohexene will be found. To separate these two reactants a simple decanter can be used on account of the very low mutual solubilities. Under industrial conditions, the hydrocarbon feed will consist of cyclohexene=cyclohexane mixtures. However, the principal reactive distillation column conguration will still work since cyclohexane will act as an inert component leaving the column at the top together with the non-reacted cyclohexene. In a similar manner, the conceptual design of a cyclohexanol splitting column can be derived from the simulated residue curves. For the lower column section one has to choose a Damkohler number above the critical value, i.e. Da 0:109; so that there is no stable node near the cyclohexanol corner. The upper part of this splitting column will be nonreactive to avoid any cyclohexanol in the distillate. Again, a simple decanter will separate water and pure cyclohexene in the top product. If one combines these two hybrid columns just mentioned, a very e cient process can be designed to reactively separate cyclohexene and cyclohexane leading to very high purities that are not easily achievable by conventional distillation. Fig. 7 shows the ow scheme of this process. This is another example of a separation process based on two coupled reactive distillation columns as suggested by Stein, Kienle, and Sundmacher (2000) for the separation of isobutene from C4 hydrocarbon mixtures.

5. Conclusions A novel process for the production of cyclohexanol by cyclohexene hydration in a hybrid reactive distillation column is suggested and analysed. Due to the very low mutual solubility of the reactants water and cyclohexene, chemical reaction and vapourliquid transfer phenomena in the column will be accompanied by liquid-phase splitting. First simulations of the simultaneous occurrence of chemical reaction, distillation and phase splitting in a batch still, and experimental investigations of the nonreactive phase behaviour have shown that liquid phase decomposition has a substantial in uence on the reactive distillation behaviour. This has to be accounted for when designing countercurrent reactive distillation columns. For this purpose, residue curve maps are used as an e cient tool to determine the stable nodes whose corresponding mixtures can be expected as column products. From the residue curve analysis, it turns out that a hybrid column is needed in order to obtain pure cyclohexanol as the bottom product. The same is true for a column to split cyclohexanol back to the educts water and cyclohexene. Most important, the ratio of the chemical reaction rate and the vaporisation rate, i.e. the Damkohler number, has to be adjusted such that a critical value is not exceeded (Da 0:109 for cyclohexanol formation) or is exceeded (Da 0:109 for cyclohexanol splitting). This could be achieved by adjustment the amount of a heterogeneous catalyst which has to be installed on the reactive column trays. The liquid-phase splitting behaviour is advantageous as an additional and simple separation mechanism that allows to bypass the azeotropes within the considered reaction system. On the other hand, it remains to be seen whether the desired Damkohler numbers can be realised within a reactive distillation column. The reaction engineering challenge here lies in the very low mutual solubilities of the two reactants that make it hard to achieve a good contact between

1520

F. Steyer et al. / Chemical Engineering Science 57 (2002) 15111520

them and the heterogeneous catalyst. But this contact will be necessary to obtain reasonable reaction rates. Depending on the catalyst used, its distribution between the two liquid phases and its intrinsic activities within the phases will vary signicantly. This will a ect the Da numbers in this system. Therefore, the focus of our future research activities will be on the experimental determination of the reaction kinetics under well-dened conditions of catalyst-liquid phase contact. Notation ai Ai Da EA H kf kf; ref Ka pisat P r R t T V xi x yi liquid phase activity of species i chemical species Damkohler number activation energy, J=mol molar liquid holdup, mol rate constant for the forward reaction, mol=(mol s) rate constant at reference conditions, mol=(mol s) chemical equilibrium constant saturation pressure of species i, Pa total pressure, Pa reaction rate, mol=(mol s) universal gas constant, =8:314 J=(mol K) time, s temperature, K vapour ow rate, mol=s average liquid phase mole fraction of component i vector of liquid mole fractions vapour phase mole fraction of component i

NC ol p r ref

number of components cyclohexanol product reactant reference state (point of lowest boiling temperature in the system) organic phase aqueous phase

References
Doherty, M. F., & Perkins, J. D. (1978). On the dynamics of distillation processesI. The simple distillation of multicomponent non-reacting, homogeneous liquid mixture. Chemical Engineering Science, 33, 281 301. Gmehling, J., Menke, J., Krafczyk, J., & Fischer, K. (1994). Azeotropic data (1st ed.). Weinheim: VCH. Gumus, Z. H., & Ciric, A. R. (1997). Reactive distillation column design with vapour=liquid=liquid equilibria. Computers & Chemical Engineering, 21(Suppl.), S983S988. Industrial Organic Chemicals, (1999). Starting materials and intermediates (1st ed.). Weinheim: Wiley-VCH. Lindstrom, P. J., & Mallard, W. G (Eds.). (2001). NIST chemistry webBook. NIST Standard Reference Database Number 69, July Release. Gaithersburg, MD: National Institute of Standards and Technology (http://webbook.nist.gov/chemistry/). Loning, S., Horst, C., & Ho mann, U. (2000). Theoretical investigations on the quaternary system n-butanol, butyl acetate, acetic acid and water. Chemical Engineering Technology, 23, 789794. Mitsui, Osamu, Fukuoka, & Yohei, (1986). Process for producing cyclic alcohol. United States Patent 4,588,846. Panneman, H. -J., & Beenackers, A. A. C. M. (1992). Solvent e ects in the liquid phase hydration of cyclohexene catalyzed by a macroporous strong acid ion exchange resin. Chemical Engineering Science, 47(9 11), 26352640. Qi, Z., Kolah, A., & Sundmacher, K. (2002). Residue curve maps for reactive distillation systems with liquid phase splitting. Chemical Engineering Science, 57(1), 163178. Stein, E., Kienle, A., & Sundmacher, K. (2000). Separation using coupled reactive distillation columns. Chemical Engineering, 107(12), 6872. Venimadhavan, G., Malone, M. F., & Doherty, M. F. (1999). A novel distillate policy for batch reactive distillation with application to the production of butyl acetate. Industrial and Engineering Chemistry Research, 38, 714722. Westerberg, A. W., Lee, J. W., & Hauan, S. (2000). Synthesis of distillation-based processes for non-ideal mixtures. Computers & Chemical Engineering, 24, 20432054. Zhicai, Y., Xianbao, C., & Jing, G. (1998). Esterication-distillation of butanol and acetic acid. Chemical Engineering Science, 53(11), 20812088.

Greek letters
i i T

R Indices ene i

relative molar organic phase holdup, mol=mol activity coe cient of component i stoichiometric coe cient of component i mole change of reaction, here = 1 dimensionless time dimensionless reaction rate

cyclohexene species

Você também pode gostar