Selectivity Engineering with Reactive Distillation for Dimerization of C

Department of Chemical Engineering, Indian Institute of Technology Bombay, Powai, .... Max-Planck-Institute for Dynamics of Complex Technical Systems...
0 downloads 0 Views 297KB Size
3024

Ind. Eng. Chem. Res. 2007, 46, 3024-3034

Selectivity Engineering with Reactive Distillation for Dimerization of C4 Olefins: Experimental and Theoretical Studies Sandip Talwalkar,† Surendra Mankar,† Amit Katariya,† Preeti Aghalayam,† Mariyana Ivanova,‡ Kai Sundmacher,‡,§ and Sanjay Mahajani*,† Department of Chemical Engineering, Indian Institute of Technology Bombay, Powai, 400076 Mumbai, India, Process Systems Engineering, Otto-Von-Guericke-UniVersity Magdeburg, UniVersita¨tsplatz 2, D-39106 Magdeburg, Germany, and Max-Planck-Institute for Dynamics of Complex Technical Systems, Sandtorstrasse 1, D-39106 Magdeburg, Germany

Kinetic analysis for the acid-catalyzed dimerization of C4 olefins was performed in a batch reactor, and a suitable rate model was developed. It was further used in the rate-based simulation model of a continuous reactive distillation (RD) column. The results obtained were compared with the experimental observations made on a pilot-scale high-pressure RD column. An important outcome of the present work is that RD can be used efficiently to enhance the selectivity of C8 dimers, thereby avoiding the formation of trimers and tetramers. The use of polar components (such as water or alcohols), as selectivity enhancers, is not necessary. Introduction The dimerization of isobutene, followed by hydrogenation of the dimerized product to form isooctane (IO), is of commercial interest, because of the environmental problems associated with the present fuel additives, such as methyl tert-butyl ether (MTBE).1,2 The dimer also finds applications as a useful intermediate and is widely used in the production of chemicals such as alkylated phenols. The dimerization can be performed in the presence of solid acid catalysts such as ion exchange resins or zeolites.3-13 Several other undesired side reactions (Figure 1), such as trimerization and oligomerization, also occur, along with dimerization; hence, the focus of the study is to enhance the selectivity toward the desired products (i.e., isomers of isooctene, commonly known as trimethyl pentenes (TMP)). As discussed in our previous work, the addition of polar components such as water, methanol, or tert-butyl alcohol (TBA) to the reaction mixture can be used to enhance the selectivity toward diisobutenes (DIB).7-10,12-15 However, this may add to the existing costs, because the added component must be separated from the final product. Moreover, TBA forms an azeotrope with DIB, which is difficult to separate.16 Hence, it may be advisable to avoid the use of polar compounds, if possible. Another school of thought is to improve selectivity through the clever choice and design of advanced multifunctional reactors, such as reactive distillation (RD). The potentially important examples of commercial interest may be found elsewhere.17,18 The underlying principle is to exploit the difference in the volatilities of IB and C8 and separate the dimers simultaneously during the course of reaction from the reactive zone of the RD column and thus minimize the consecutive oligomerization reactions that are initiated from DIB and other C8 olefins. Moreover, such an approach has also been proved effective for the proper utilization of heat in the liquid-phase * To whom correspondence should be addressed. Tel.: +91-2225767246. Fax: +91-22-25726895. E-mail: [email protected]. † Department of Chemical Engineering, Indian Institute of Technology Bombay. ‡ Process Systems Engineering, Otto-von-Guericke-University Magdeburg. § Max-Planck-Institute for Dynamics of Complex Technical Systems.

Figure 1. Reaction network of oligomerization of IB3 (TEB denotes tetraisobutenes).

oligomerization of C4 and C5 olefins.19 The feasibility of this approach has been theoretically evaluated for the dimerization of IB in the earlier work in our research group.20,21 The simulations showed that RD with the proper choice of design and operating parameters can significantly enhance the selectivity. The reaction kinetics used in these studies is the one developed by Honkela and Krause,13 which was generated in the presence of the polar component (water-TBA). To predict the performance of RD without the polar component, the kinetics was extrapolated to dry conditions by setting the concentration of water-TBA to zero. Such an assumption may not be valid for the reliable prediction of the reactor performance. The reaction kinetics in the presence and absence of polar compounds are likely to be significantly different. Hence, there is a need to develop a separate kinetic model that is valid in the absence of polar components. The previous work on the kinetics of this reacting system has been reviewed extensively in our previous work.14 In most of the studies, including our own work, catalyst inhibitors have been used to enhance the selectivity. However, barring a few cases,6,22 the kinetics over a dry catalyst has not received much attention, probably because of the realization of poor selectivity toward the dimers in the conventional reactor system. The objective of the present work is to evaluate the feasibility of RD for a possible improvement in the selectivity, especially in the absence of any selectivity-enhancing components and validate qualitatively, if not quantitatively, the simulation results in earlier work. Our starting material is a C4 stream mixture, instead of pure IB, and, hence, it should be noted that, along with the formation of DIB, there exists a possibility of cross

10.1021/ie060860+ CCC: $37.00 © 2007 American Chemical Society Published on Web 04/18/2007

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3025

Figure 2. Typical chromatograms for the reaction mixture.

Table 1. Composition (% mol) of the Feed C4 Mixture on a Solvent-Free Basis component propylene isobutane n-butane butene-1 isobutene trans-butene-2 cis-butene-2 heavier compounds

amount (% mol) 0.048 3.15 8.10 25.48 48.10 8.67 4.48 1.97

dimerization of IB with other linear butenes.12 A kinetic model is developed for such a system under nonpolar conditions, based on the laboratory batch reactor data. Batch and continuous experiments have been performed on a pilot-scale RD column. To the best of our knowledge, the experimental results on C4 dimerization in a RD column are being reported for the first time in the open literature, through this work.

Table 2. Physical Properties of the Dry Ion-Exchange Resin Catalyst T-63 Used in This Work and Its Comparison with Amberlyst-15 Value property

Amberlyst-15

T-63

crosslinking total capacity porosity specific surface area pore diameter temperature stability screen size

15% 4.9 mequiv/g 49% 45 m2/g 265 Å 120 °C 16-50 US mesh

15% 4.9 mequiv/g 40% 35 m2/g 400 Å 120 °C 16-50 US mesh

Experimental Section Materials. IO (99.5%) was obtained from Merck, Ltd., India. DIB and the C4 stream of required composition (Table 1) were obtained from Hardillia Schenctady Ltd., Mumbai, India, and used to perform both all the kinetics and RD runs. Thermax Ltd., India, supplied Tulsion T-63, which was used as as catalyst.

3026

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

Figure 4. Effect of particle size (([) 16-50 US mesh, (2) 50-70 US mesh, and (b) 70-100 US mesh). Catalyst, T-63; T ) 343 K; catalyst loading, 0.0116 w/w of IB; C4/IO ) 3 w/w.

Figure 3. General course of the reaction. Catalyst, T-63; T ) 363 K; catalyst loading, 0.0116 w/w of isobutene; C4/IO ) 3 w/w (IO denotes iso-octane).

This catalyst was dried at 80 °C at 50 mm Hg for ∼12 h. The physical properties of the catalysts are given in Table 2 and are compared with the commonly used Amberlyst-15 catalyst (from Rohm and Haas, Philadelphia, PA). Apparatus and Procedure. A stainless steel autoclave (Parr Instrument Company, USA) with a capacity of 3 × 10-4 m3, which was equipped with an on-line temperature and speed monitoring facility was used to conduct all the batch reactions. The desired quantities of the catalyst (0.2-0.8 g), C4 (80-100 g), and solvent (IO, 20-50 g) were charged to the reactor and the reaction mixture was heated to the desired temperature with slow stirring. As the reaction temperature was attained, the speed of agitation was increased up to the desired level and the corresponding time was regarded as the zero reaction time. The liquid samples were withdrawn at different time intervals and analyzed to study the kinetics of the reaction. All the runs were conducted at autogenous pressure and the initial pressure in the batch runs was in the range of 6-9 atm, depending on the temperature of the reaction. The pressure decreases gradually as the reaction proceeds. Analysis. The reactants and products were analyzed using a gas chromatograph (GC-MAK-911) that was equipped with a flame ionization detector (FID). A 25-m-long capillary column BP-5 (SGE, Australia) was used to separate the different components in the product mixture, using iso-octane as an

internal standard. The column temperature was maintained at 333 K isothermally for the first 5 min and then was increased, at a rate of 30 K/min, up to 523 K for 20 min. A typical chromatogram of the reaction mixture, showing all the important species involved, is given in Figure 2. The components that elute after the triisobutenes (TIB) are called oligomers. The product mixture was fractionated by distillation under vacuum and used to calibrate the trimers and oligomers. The various components in the product mixture and the separated products were characterized either by authentic samples and/or by gas chromatography-mass spectroscopy (GC-MS). The column does not separate C4 compounds, and, hence, the C4 stream was analyzed separately on a gas chromatograph (GC-MAK-911) that was equipped with an FID device. A 12-m-long 25% Saebaconitrile packed column (Chromatopak, India) was used to separate various hydrocarbons in the C4 stream. The column temperature was maintained at 303 K. The source of the isobutylene is C4 and it contains considerable amounts of 1and 2-butene; therefore, the cross dimerization of isobutene with linear butene is likely to occur and this mixture of C8 not only contains diisobutylenes but also the cross dimers (c-DIM). The peaks of cross dimers were identified by comparing the chromatograms of the product obtained on the dry catalyst in two different cases, i.e., with pure isobutene and with C4 as a source of isobutene. The additional peaks with C4 as feed were considered as peaks of cross dimer and the response factor for these cross dimers was assumed to be the same as that of diisobutylenes. In the case of RD operation, the samples were collected in a high-pressure sampling bomb in which iso-octane was initially added as a standard. The bomb was weighed, cooled, and connected to the gas chromatograph (GC) sampling valve and the gases from the bomb were injected in the GC for the analysis of C4 gases. The liquid from the bomb was analyzed independently, to determine the amount of various products formed in the reaction. Calculations for Conversion and Selectivity. The conversion of IB in the batch reactor was calculated based on the various products formed in the reactions, as

XIB )

2nDIB + 3nTIB + nc-DIM + 4nTEB nIB,0

(1)

(TEB denotes tetraisobutenes) and the selectivity toward C8 is

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3027

Figure 5. Effect of temperature on the product profiles ((O) 323 K, (2) 343 K, and (0) 363 K). Catalyst, T-63; catalyst loading, 0.0116 w/w of IB; C4/IO ) 3 w/w.

calculated as

S)

2nDIB + nc-DIM 2nDIB + 3nTIB + nc-DIM + 4nTEB

(2)

Batch Kinetics: Results and Discussion Prior to conducting the kinetic runs, few representative batch RD runs were performed to get an idea about the typical concentration and temperature region associated with the reactive zone of the RD column. Because isobutene and other C4 compounds are much more volatile than the other components, the reactive zone that is placed in the rectifying section is rich in C4 and the temperature is in the range of 333-363 K. Hence, the experiments on batch kinetics were performed under these conditions. The data points in the initial period, when the reaction mixture is rich in isobutene, were considered for the estimation of the model parameters. General Course of the Reaction. Figures 3a and b show typical concentration profiles of the various products, with respect to time. As the reaction proceeds, the concentration of various products increases and the selectivity to DIB/C8 decreases with an increase in IB conversion. As mentioned previously, we define the zero reaction time as the time at which the desired temperature is attained. Hence, in all the kinetic runs, we observe a small extent of reaction at time zero. The extent of reaction during this heat-up period was determined to be relatively higher at higher catalyst loadings and temperatures.

The same values of initial concentration of various components were used as initial points for parameter estimation, using a batch reactor model. The moisture has a crucial role in the selectivity of DIB/C8 as water, being a polar compound, covers the active catalytic sites, thereby reducing the relative rates of the oligomerization reactions.14 Therefore, during the charging of the reactants, adequate care was taken so that no external moisture from the humid atmosphere can leak into the reacting system. If the moisture is present in the reactor, traces of TBA, which is formed by hydration of IB, are observed in the gas chromatogram. The absence of the TBA peak in our samples at all times confirms the complete moisture-free environment for the reaction. Mass-Transfer Effects. To ensure the absence of external mass-transfer resistance across the solid/liquid interface, the reactions were performed at sufficiently high speed of agitation, i.e., 1700 rpm. Moreover, as shown in Figure 4, the catalyst size does not have any significant effect on the reaction rate in the region of interest, i.e., in the initial period of the reaction. All the kinetic runs were performed with the particle size in the range of 70-100 mesh. Effect of Temperature. The effect of temperature was studied in the range 323-363 K. Figure 5 shows the effect of temperature on the concentration profiles for the various products. As expected, the overall rate of oligomerization increases as the temperature increases, at the cost of decreasing

3028

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

Figure 6. Effect of catalyst loading on product profiles ((O) 0.0046 g/(g of IB charged), (2) 0.0116 g/(g of IB charged), and (0) 0.0185 g/(g of IB charged)). Catalyst, T-63; C4/IO ) 3 w/w.

selectivity toward DIB/C8, as the side reactions are more sensitive to changes in temperature. Effect of Catalyst Loading. The effect of catalyst loading was studied over the range of 0.0086-0.0185 g/(g of IB charged). Figure 6 shows the effect of catalyst loading on the concentration profiles for the various products. As expected, the overall rate of the reaction increases with the catalyst loading. It may be noted that the catalyst loading used in these runs is significantly lower than that used in the case of reactions with polar compounds14 for the obvious reason that the reaction is much faster over the dry catalyst in the absence of the polar component. Effect of IB Concentration. The effect of initial concentration of IB on the rate of reaction was studied and, as expected, the rate of the reactions increases with increasing IB concentration, as shown in Figure 7. Kinetic Modeling Reaction Scheme. The possible routes to form different oligomers of IB are shown in Figure 1. However, in our earlier study,14 we have shown that no formation of TIB or TEB is observed when the reaction is performed starting with pure DIB/ C8 over the parameter range of interest. Hence, it allows one to eliminate several other reaction pathways. Thus, the reactions to be considered in the kinetic model are as follows:

IB + IB f DIB

(reaction 1)

DIB + IB f TIB

(reaction 2)

TIB + IB f TEB

(reaction 3)

IB + 2-Bu f c-DIM

(reaction 4)

1-Bu f 2-Bu

(reaction 5)

As mentioned previously, because the reaction in the reactive zone of the column occurs in a C4-rich region (the concentration of C4 is >95% mol/mol), one can use a simplified concentrationbased PH (pseudo-homogeneous) model, as given in Table 3. Reaction 5 was considered irreversible, because the reverse rate constant obtained by regression was negligibly small. The depletion in the concentration of butene-1 supports the fact that isomerization of butene-1 to butene-2 occurs under the reaction conditions. This has been proved experimentally by Petrus et al.35 in the context of sec-butanol synthesis from n-butenes over an ion-exchange resin. Parameter Estimation. The component mass balance for the batch reactor are given by

dni dt

NR

) Mcat

υi,jrj ∑ j)1

The temperature dependency of the rate constant can be

(3)

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3029

Figure 7. Effect of concentration of IB on product profiles ((O) 24.25 mol %, (2) 32.76 mol %, and (0) 37.54 mol %). Catalyst: T-63; T ) 363 K; catalyst loading, 0.5 g.

Table 3. Kinetic Expressions for the Reacting System reaction

rate equation

1 2 3 4 5

r1 ) k1xIB2 r2 ) k2xIBxDIB r3 ) k3xIBxTIB r4 ) k4xIBx2-Bu r5 ) k5x1-Bu

Table 4. Values of Parameters and the Objective Function Φ for the Proposed Model parameter

expressed, using the Arrhenius equation, as

( )

ki ) ki,0 exp

-Ei,0 RT

(4)

The overall objective function Φ for optimization to minimize the difference between measured and predicted values is given by

Φ)

(xi,calculated - xi,measured)2 ∑ all samples

(5)

For optimization, a sequential quadratic programming (SQP) algorithm from the NAG library was used in the simulation environment DIVA.23 Estimated values of the parameters for the kinetic model are given in Table 4, while the comparison between measured and calculated concentrations for the developed model is shown in Figure 8. Note that the kinetic model and the parameters reported here are valid only in the case of relatively large C4 concentrations and, hence, can be used with sufficient confidence level to predict the performance of the

value

k1,0 E1,0 k2,0 E2,0 k3,0 E3,0 k4,0 E4,0 K5,0 E5,0

3.01 × 1013 mol kg-1 s-1 89.34 kJ/mol 3.38 × 104 mol kg-1 s-1 26.70 kJ/mol 6.45 × 104 mol kg-1 s-1 31.0 kJ/mol 1.32 × 108 mol kg-1 s-1 57.61 kJ/mol 5.63 × 1011 mol kg-1 s-1 86.09 kJ/mol

Φ

0.031

RD column used in the present case. The sensitivity, with respect to the kinetic parameters, was studied using the sensitivity analysis method in DIVA.23 The deviation from the base case was determined by disturbing the parameters by (10% from the base case values. It is observed that the conversion of IB and the selectivity toward DIB/C8 are sufficiently sensitive to all the modeled parameters. Reactive Distillation Experimental Setup. All the reactions were performed using the experimental setup shown in Figure 9, and the design details are given in Table 5. A hybrid column with a total height of 3 m, which consists of a reactive rectifying section (1 m) packed

3030

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 Table 5. Reactive Distillation Column Details parameter

value

column diameter, ID column height height of reactive section total catalyst loading height of nonreactive section number of theoretical stages in the reactive zone number of nonreactive stages operating pressure feed pump (CRD) feed location (CRD) feed pump rating reboiler holdup condenser-accumulator holdup

5 cm 3m 1m 30 g 2m 3 16 10 atm diaphragm type bottom of reactive zone: stage 4 from top 0-3 L/h 2L 0.3 L

Table 6. Operating Parameters and Conversion and Selectivity for the Batch Reactive Distillation (BRD) Runs

Figure 8. Comparison between modeled and measured values.

with KATAPAK-S and a nonreactive stripping section (2 m) packed with EVERGREEN Hyflux packing, was used. In the present reacting system, the reactant is more volatile than the product(s) and, hence, it was advisable to place the reactive zone in the rectifying section.24 The catalytic packing was loaded with ion-exchange resin catalyst (T-63). The column was lagged with glass wool to prevent the heat losses to the surroundings. The column was equipped with sampling points, temperature sensors, and feed points at different locations spaced apart by 0.3 m. The heat to the reboiler was provided by hot circulating oil that is heated electrically in a separate vessel. Chilled water was circulated through the condenser to remove heat. In the case of continuous runs, the feed was supplied at a constant rate, with the help of a diaphragm pump. In all cases, the feed was introduced at the bottom of the reactive zone. Batch Reactive Distillation (BRD). Prior to performing the continuous runs, which are relatively more time-consuming and require a large amount of chemicals, it is advisable to first perform the runs in a batch mode in the same setup. This helps one to quickly evaluate the feasibility of RD for a given reaction and also to obtain an appropriate initial condition for the startup of the continuous run. A known amount of C4 was charged to the reboiler initially and the heating was started at a desired rate. The temperatures and compositions at several locations in the column were measured with respect to time. The column holdup was assumed to be negligible, compared to the reboiler holdup, and conversion and selectivity were calculated based on the reboiler composition and the holdup. The results on the effect of various parameters are given in Table 6. The reproducibility of the results was checked and the error in the results was determined to be within 3%-5%. The selectivities to C8 obtained are remarkably high, compared to that obtained in the conventional reactor systems with a dry catalyst, i.e., in the absence of polar component, as shown in Table 7. The typical column concentration profiles are shown in Figure 10. The stage number was calculated by considering NSTM for reactive and nonreactive packings as 3 and 8, respectively, based on the data supplied by respective suppliers. One can notice that the reactive zone (stage numbers 1-3) contains a negligible concentration of DIB/C8, which is the reason a considerable improvement in the selectivity is realized. All the batch and continuous runs were performed on the same catalyst, with ∼150 h of run time, and it was determined that the catalyst does not deactivate. This, we believe, is due to simultaneous separation of the heavier

run

batch time (h)

reboiler duty (kW)

conversion (%)

selectivity (%)

1 2 3 4 5 6 7

4 4 11 4 6 4 7

1.9 1.9 1.9 2.3 2.3 2.5 2.5

52.77 51.20 99.58 74.99 96.91 75.36 97.53

86.00 86.65 85.71 85.10 84.53 85.86 83.84

products from the reactive rectification zone, offering a relatively clean reaction environment for the reaction. Continuous Reactive Distillation (CRD). The experimental setup was operated in a continuous mode by feeding the reactant C4 stream at a desired location and withdrawing the bottom product, continuously. The column was initially operated in a batch mode, and after the desired concentration and temperature profiles were obtained, a continuous feed to the column was introduced. The product withdrawal was also started simultaneously. The samples were withdrawn at different time intervals and the temperatures were measured until there was no further change in these parameters, i.e., when the steady state was attained. After this, the run was continued for a few more hours to ensure the attainment of the steady state. The experiments were performed at various reboiler duties, and the results are presented in Table 8. The selectivities to C8 obtained are relatively higher than that obtained in BRD. This is anticipated as, in BRD, the concentration of DIB/C8 in the reboiler and on the stages slowly increases with time, which adversely affects the selectivity whereas in the CRD runs DIB/C8 are continuously removed. It can be observed that the selectivity increases with as the reboiler duty increases. It is known that the interplay of reaction and distillation must be well-understood while using RD for selectivity enhancement. The system of interest is a typical example of the single-reactant, series-parallel reaction system wherein the volatility differences between the reactant and all the products are very high (RIB-DIB > 20). In such case, the reboiler duty and per-stage Damko¨hler number (Da) have a crucial role in determining the selectivity at a given conversion. For a non-azeotropic system such as the present one, the attainable selectivity in ideal cases is close to 100%, and there are no thermodynamic and kinetic constraints.24 An increase in the reboiler duty and a decrease in the per-stage Da value would help one to achieve the desired enhanced performance. We varied the reboiler duty and, as shown in Table 8, the results are in concurrence with the theory. Modeling and Simulation. The performance of the CRD column was predicted with the help of a rate-based model. The

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3031

Figure 9. Schematic of the experimental setup used for reactive distillation (RD). (The top two sections are reactive and are packed with KATAPACK-S.)

Figure 11. Comparison of experimental and simulation for run 1 in Table 8. The shaded region indicates the reactive zone. Figure 10. Typical column liquid concentration profiles for batch reactive distillation (BRD).

model uses the kinetics developed in the present work. Because of the fact that the intrinsic reaction over the dry catalyst is extremely fast, the mass-transfer resistance across the solid/ liquid interface must be considered.28 Attempts to apply a twophase (V-L) equilibrium stage model without considering the mass-transfer resistances across the solid/liquid interface failed because the predicted conversion and selectivity values were too high, compared with the experimental results. Hence, a rigorous three-phase nonequilibrium (NEQ) model has been developed, considering the effect of mass-transfer resistances. The purpose is to compare the performance and behavior with that obtained from the experiments. We refer to Powers et al.30 for the detailed model implementation and computational aspects. In the case of NEQ models, the hardware design specification (column diameter, tray or packing type and geometry, etc.) is mandatory. The column design is taken from the existing experimental setup described previously. The packing selected for the reactive and nonreactive (rectifying and stripping) sections are KATAPAK-S and Sulzer-BX, respec-

tively. Sulzer-BX has the similar characteristics as that of Evergreen HYFLUX packing used in the experiments. The hydraulic and mass-transfer correlations for vapor-liquid resistances for these packing are obtained from Rocha et al.31,32 and Kolodziej et al.,33 respectively. The correlations for the calculation of solid-liquid mass transfer are obtained from Xu et al.34 For the vapor-liquid equilibrium, the UNIQUAC model was used to determine the activity coefficients of the various components involved. The corresponding binary interaction parameters calculated by UNIFAC, using the Aspen database,27 are given in Table 9. The vapor-phase nonideality was considered and the fugacity was calculated using the Peng-RobinsonStryjek-Vera method.20 For simplicity, all the inert components (propylene, n-butane, isobutane, and heavier) are treated as isobutane. The packed column is divided into several segments, each of which acts as a NEQ stage. The stage equations are the equations for mass and energy balances for the individual phases, in which mass- and heat-transfer rates are also included. The bulk variables (compositions, flow rates, molar fluxes, energy fluxes, temperatures) are different from that of interface

3032

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

Table 7. Experimental Conversions and Selectivities for Dimerization of Isobutene: Performance of Different Reactors

a

Sr. No

reactor type

polar component?

1 2 3 4 5 6 7 8

CSTR batch batch batch RD fixed bed fixed bed catalytic membrane continuous RD

conversion (%)

selectivity (%)

ref(s)

∼100 95 60 85 98-93 60-42 80-22 93

[12, 13] [14] [24] present worka [15] [25] [26] present worka

∼35 60 70 ∼100 28-62 25-55 30-98 73

yes yes no no yes no no no

The results given represent the experimentally obtained state of best performance.

Table 8. Operating Parameters for CRD and the Experimental and Predicted Values of Conversions and Selectivities Conversion (%)

Selectivity (%)

Temperature at 3 m (K)

Reboiler Temperature (K)

run

feed flow rate (kg/h)

reboiler duty (kW)

measured distillate flow rate (kg/h)

distillate flow rate used in simulation (kg/h)

expt

simul

expt

simul

expt

simul

expt

simul

1 2 3

0.47 0.47 0.47

1.9 2.3 2.5

0.303 0.29 0.289

0.3130 0.3045 0.3085

49.54 55.56 59.69

48.90 51.09 51.58

90 92 93

90.68 91.4 91.9

333.35 335.55 342.45

339.08 339.38 339.48

434.4 429.45 447.05

434.29 429.24 447.3

Table 9. Binary UNIQUAC Interaction Parameters Used for Continuous Reactive Distillation (CRD) Simulationa

IB DIB TIB c-DIM isobutane 1-butene 2-butene TEB Q R a

IB

DIB

TIB

c-DIM

isobutane

1-butene

2-butene

TEB

0 -51.44 -45.06 -50.51 -37.69 -56.98 -60.59 -44.92

45.89 0 -58.29 -60.42 -47.29 -31.13 -51.75 -58.19

38.99 52.69 0 -58.67 -54.33 -12.87 -38.76 -60.17

44.89 54.71 53.07 0 -49.17 -27.34 -49.17 -58.87

28.54 41.54 48.81 43.52 0 26.5 30.24 48.21

51.49 26.28 6.99 22.41 -39.23 0 -55.57 7.93

54.87 46.55 33.92 44.07 -38.49 50.04 0 34.43

38.90 52.60 54.48 53.24 -53.72 -13.73 -39.29 0

2.68 2.92

4.92 5.62

6.88 8.31

4.92 5.61

2.77 3.15

2.56 2.92

2.56 2.92

9.04 11.01

Data taken from ref 27.

variables. Equilibrium is assumed to be only at the interface and temperatures of vapor and liquid streams are not identical. The condenser and reboiler are treated as equilibrium stages. Steady-state simulations, with the help of the developed NEQ model, were conducted for the design and operating parameters given in Table 8. The NEQ model equations were implemented in the dynamic simulator DIVA.29 DIVA uses an equationoriented approach by solving all differential and algebraic equations (DAE) simultaneously. This comes with an in-built package for continuation and stability analysis for DAE systems. Initially, the column height in the NEQ model was divided in the same number of slices as the number of equilibrium stages, i.e., 19 (3 slices in the reactive section and 16 slices in the nonreactive stripping section). The results from the equilibriumstage model were used as initial guesses to the nonlinear algebraic equation solver. The guessed values for bulk and interface variables were assumed to be the same. The number of segments in each section of the column was increased such that there was no further change in the column profiles, the conversion of isobutylene, and the selectivity of the dimer. Because of the fact that no data or model are published for the estimation of the solid-liquid mass-transfer coefficient for the Katapack-S packing, we used the available correlation in the literature34 and multiplied the value by a gain factor chosen as explained below. It was observed that the simulation results were very sensitive to a few input parameters, especially the distillate flow rate and mass-transfer coefficients of IB and C8. The distillate flow rate and the mass-transfer coefficients (i.e., the gain factor) were varied in the simulations, such that there was a satisfactory match between experimental and simulation results. The distillate flow rates thus obtained are compared with the experimen-

tally measured values. As shown in Table 8, the measured values lie within 10% of the values obtained by simulation. Note that the error in material balance closure is ∼5%. Figure 11 shows the comparison of simulation and experimental results for both concentration and temperature profiles in the column for run 1. The comparison for the other two representative runs is also satisfactory. For all the runs and along the column height, the solid-liquid mass-transfer coefficient varies from 0.6 × 10-6 to 3 × 10-6 m/s for IB and, 0.01 × 10-6 to 0.2 × 10-6 m/s for the C8 compounds. The values have the same order of magnitude as that obtained for similar compounds using standard correlations such as that by Wilson and Genkoplis.36 Note that, although the agreement between experimental and predicted results is satisfactory, the authenticity of the mass-transfer coefficients, under the conditions of interest, must be confirmed through independent experiments. Detailed parametric studies that use the validated NEQ simulation model to arrive at the best-possible conditions that give desired yields is the subject of our future investigations. Conclusions We have performed detailed kinetic analysis for the dimerization of isobutene over a dry ion-exchange resin and proposed an appropriate rate model that will be useful for the design and simulation of reactive distillation (RD) units. It has been shown that the use of RD is beneficial, in terms of the improvement in selectivity toward the intermediate products DIB/C8. The use of polar components such as water or alcohols may not be necessary and one can get significant yields via RD alone. A three-phase model is used to explain the experimental results, and the predictions are determined to be very sensitive to the

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3033

solid-liquid mass-transfer coefficients used. The model may be further used to arrive at the optimal configuration that gives the desired yield of dimers. Acknowledgment The financial support for this work from the Volkswagen Foundation in Germany (VW Stiftung Project “Coupling of Chemical Reactions in a Reactive Distillation Process”, No. AZ I/79515) is gratefully acknowledged. Authors also acknowledge the Department of Science and Technology (DST), India for the financial support to build the high-pressure reactive distillation facility at IIT-Bombay. The assistance by S. Srinivas (Research Scholar, IIT-Bombay) in parameter estimation studies is gratefully acknowledged. Nomenclature RIB-DIB ) relative volatility of isobutene, with respect to diisobutene Da ) Damkohler number Ei,0 ) activation energy for reaction i (kJ/mol) ki,0 ) rate constant for reaction i (mol s-1 kgcat-1) MCAT ) mass of catalyst (kg) ni ) number of moles of component i (mol) ni,0 ) initial number of moles of component i (mol) NR ) number of reactions rj ) rate of reaction j (mol/s) S ) selectivity towards dimer xi ) liquid molar fraction of component i X ) conversion of isobutene υi,j ) stoichiometric coefficient for component i and reaction j Φ ) objective function for optimization AbbreViations 2-Bu ) cis-butene-2 + trans-butene-2 1-Bu ) butene-1 BRD ) batch reactive distillation CRD ) continuous reactive distillation DIB ) diisobutenes c-DIM ) cross dimer of isobutene and butene-2 IB ) isobutene IO ) isooctane MTBE ) methyl tert-butyl ether NEQ ) nonequilibrium PH ) pseudo-homogeneous RD ) reactive distillation TBA ) tert-butyl alcohol TEB ) tetraisobutenes TIB ) triisobutenes TMP ) trimethyl pentene Literature Cited (1) Di Girolamo, M.; Tagliabue, L. MTBE and Alkylate Coproduction: Fundamentals and Operating Experience. Catal. Today 1999, 52, 307. (2) Executive Order D-52-02 by the Governor of the State of California, March 15, 2002. (3) Kolah, A. K.; Qi, Z.; Mahajani, S. M. Dimerised isobutene: An alternative to MTBE. Chem. InnoVation 2001, 31, 15. (4) Marchionna, M.; Di Girolamo, M.; Patrini, R. Light olefin dimerization to high quality gasoline components. Catal. Today 2001, 65, 397. (5) O’Conner, C. T.; Kojima, M.; Schumann, W. K. The Oligomerization of C4 Alkenes over Cationic Exchange Resins. Appl. Catal. 1985, 16, 193. (6) Haag, W. O. Oligomerization of Isobutylene on Cation Exchange Resins. Chem. Eng. Prog., Symp. Ser. 1967, 63, 140. (7) Bowman, W.; Stadig, W. Dimerization of Isobutene. U.S. Patent 4,100,220, July 11, 1978.

(8) Evans, T. I.; Karas, L. J.; Rameswaran, R. Selective olefin oligomerization. U.S. Patent 6,376,731, April 23, 2002. (9) Di Girolamo, M.; Lami, M.; Marchionna, M.; Pescarollo, E.; Tagliabue, L.; Ancillotti, F. Liquid-Phase Etherification/Dimerization of Isobutene over Sulphonic Acid Resins. Ind. Eng. Chem. Res. 1997, 36, 4452. (10) Vila, M.; Cunill, F.; Izquierdo, J.; Gonzalez, J.; Hernandez, A. The role of by products formation in methyl tert-butyl ether synthesis catalysed by a macro porous acidic resin. Appl. Catal., A 1994, 117, L99. (11) Stine, L. O.; Muldoon, B. S.; Gimre, S. C.; Frame, R. R. Process for oligomer production and saturation. U.S. Patent 5,895,830, April 20, 1999. (12) Honkela, M.; Krause, A. O. I. Influence of polar components in the dimerization of isobutene. Catal. Lett. 2003, 87, 113. (13) Honkela, M.; Krause, A. O. I. Kinetic Modeling of the Dimerization of Isobutene. Ind. Eng. Chem. Res. 2004, 43, 3251. (14) Talwalkar, S.; Chauhan, M.; Aghalayam, P.; Qi, Z.; Sundmacher, K.; Mahajani, S. Kinetic Studies on the Dimerization of Isobutene with Ion-Exchange Resin in the Presence of Water as Selectivity Enhancer. Ind. Eng. Chem. Res. 2006, 45, 1312. (15) Ouni, T.; Honkela, M.; Kolah, A.; Aittamaa, J. Isobutene dimerization in a miniplant-scale reactor. Chem. Eng. Process. 2006, 45, 329. (16) Wang, J.; Sahay, N.; Loescher, M.; Vichaliak, M. Separation of tertiary butyl alcohol from diisobutylene. U.S. Patent 6,863,778, March 8, 2006. (17) Sharma, M. M.; Mahajani, S. M. Industrial Applications of Reactive Distillation. In ReactiVe Distillation: Status and Future Directions; Sundmacher, K., Kienle, A., Eds.; Wiley-VCH: Weinheim, Germany, 2003. (18) Hiwale, R.; Bhate, N. V.; Mahajan, Y.; Mahajani, S. M. Commercial applications of reactive distillation: Recent trends. Int. J. Chem. React. Eng. 2002, 2, R1. (19) Smith, L.; Houston, B.; Jones, E.; Vichaliak, M. Oligomerization process. U.S. Patent 5,003,124, 1991. (20) Kamath, R.; Qi, Z.; Sundmacher, K.; Aghalayam, P.; Mahajani, S. Process Analysis for Dimerization of Isobutene by Reactive Distillation. Ind. Eng. Chem. Res. 2006, 45, 1575. (21) Kamath, R.; Qi, Z.; Sundmacher, K.; Aghalayam, P.; Mahajani, S. Comparison of Reactive Distillation with Process Alternatives for the Isobutene Dimerization Reaction. Ind. Eng. Chem. Res. 2006, 45, 2707. (22) Alcantara, R.; Alcantara, E.; Canoira, L.; Franco, J.; Herrera, M.; Navarro, A. Trimerization of isobutene over Amberlyst-15 catalyst. React. Funct. Poly. 2000, 45, 19. (23) Mangold, M.; Kienle, A.; Gilles, E. D.; Mohl, K. D. Nonlinear computation in DIVAsmethod and applications. Chem. Eng. Sci. 2000, 55, 441. (24) Agarwal, V.; Thotla, S.; Mahajani, S. M. Selectivity engineering with reactive distillation: determination of attainable region. 8th Distillation and Absorption, 2006; IchemE Symp Sec. No. 152, 2006; p 73. (25) Mantilla, A.; Tzompantzi, F.; Ferrat, G.; Lopez-Ortega, A.; Alfaro, S.; Gomez, R.; Torres, M. Oligomerization of isobutene on sulfated titania: Effect of reaction conditions on selectivity. Catal. Today 2005, 107-108, 707. (26) Fritsch, D.; Randjelovic, I.; Keil, F. Application of a forced-flow catalytic membrane reactor for the dimerization of isobutene. Catal. Today 2004, 98, 295. (27) Aspen Technologies, Inc. Aspen Version 11.1, Aspen Plus, Cambridge, MA, 2001. (28) Huange, C.; Ng, F.; Remple, G. Application of catalytic distillation for the aldol condensation of acetone: the effect of the mass transfer and kinetic rates on the yield and selectivity. Chem. Eng. Sci. 2000, 55, 5919. (29) Kroner, A. P.; Holl, W.; Marquardt; Gilles, E. D. DIVA: An open Architecture for Dynamic Simulation. Comput. Chem. Eng. 1990, 14 (11), 1289. (30) Powers, M. F.; Vickery, D. J.; Arehole, A.; Taylor, R. A non-equilibrium stage model of multicomponent separation processessV. Computational method for solving model equations. Comput. Chem. Eng. 1998, 12 (12), 1229. (31) Rocha, J. A.; Bravo, J. L.; Fair, J. R. Distillation Columns Containing Structured Packings: A Comprehensive Model for Their Performance. 1. Hydraulic Model. Ind. Eng. Chem. Res. 1993, 32 (4), 641. (32) Rocha, J. A.; Bravo, J. L.; Fair, J. R. Distillation Columns Containing Structured Packings: A Comprehensive Model for Their Performance. 2. Mass-Transfer Model. Ind. Eng. Chem. Res. 1996, 35 (5), 1660. (33) Kolodziej, A.; Jaroszynski, M.; Bylica, I. Mass transfer and hydraulics for KATAPAK-S. Chem. Eng. Process. 2004, 43 (3), 457.

3034

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

(34) Xu, Y.; Flora, T. N.; Rempel, G. L. Comparison of a Pseudohomogeneous Non-equilibrium Dynamic Model and a Three-phase Nonequilibrium Dynamic Model for Catalytic Distillation. Ind. Eng. Chem. Res. 2005, 44, 6171. (35) Petrus, L.; De Roo, R. W.; Stamhuis, E. J.; Joosten, G. E. H. Kinetics and equilibria of the hydration of linear butenes over a strong acid ion-exchange resin as catalyst. Chem. Eng. Sci. 1986, 41, 217.

(36) Wilson, E. J.; Geankoplis, C. J. Liquid mass transfer at very low Reynolds numbers in packed beds. Ind. Eng. Chem. Fundam. 1966, 5, 9.

ReceiVed for reView July 4, 2006 ReVised manuscript receiVed February 28, 2007 Accepted February 28, 2007 IE060860+