ETBE Synthesis via Reactive Distillation. 1. Steady-State Simulation

Ind. Eng. Chem. Res. 1997, 36, 1855-1869

1855

ETBE Synthesis via Reactive Distillation. 1. Steady-State Simulation and Design Aspects Martin G. Sneesby,*,† Moses O. Tade´ ,† Ravindra Datta,‡ and Terence N. Smith† School of Chemical Engineering, Curtin University of Technology, Perth, Western Australia, and Department of Chemical and Biochemical Engineering, University of Iowa, Iowa City, Iowa 52242

Ethyl tert-butyl ether (ETBE) is an alternative gasoline oxygenate that combines the blending properties of methyl tert-butyl ether (MTBE) and the renewability of ethanol. Technologically, the best means of synthesis utilizes reactive (or catalytic) distillation to maximize hydrocarbon conversion and energy efficiency while simultaneously producing a high-purity ether product. Mathematical models of reactive distillation are based on the conventional distillation process with supplementary equations added to model the reactions present. Ether-alkene-alcohol systems are highly nonideal in the liquid phase so that careful selection of physical property routines is required to ensure satisfactory simulation results. Column simulations performed here using both Pro/II and SpeedUp show excellent agreement with previously published experimental data for a MTBE system and also agree well with each other for both MTBE and ETBE systems. A homotopy analysis was performed on the tuned simulation models to determine the effects of key design and operating variables on column performance and, subsequently, to develop a design method for reactive distillation columns. Some unusual behavior was identified in ETBE reactive distillation columns compared with either MTBE columns or conventional distillation. Introduction Changing worldwide regulations are encouraging the addition of oxygenates to gasoline sold in heavily urbanized areas to reduce emissions of carbon monoxide and unburned hydrocarbons in an attempt to combat smog and ground level ozone. The high octane rating of many oxygenates can also be utilized to eliminate leaded octane enhancers, such as tetramethyllead (TML) and tetraethyllead (TEL), from gasoline blends. To date, methyl tert-butyl ether (MTBE) and ethanol have been the most widely used oxygenates. MTBE appears to offer the best combination of oxygen content, low Reid vapor pressure (RVP), high octane, high energy content, and low cost, but ethanol has been used in gasoline for many years and has attracted particular interest as an environmentally friendly alternative to fossil fuels, as it can be produced from biomass. Many governments also offer ethanol subsidies to offset the cost differential with MTBE. Ethyl tert-butyl ether (ETBE) has emerged more recently as a potential oxygenate and offers the advantages of the blending characteristics of MTBE and the renewability of ethanol. Compared with MTBE, ETBE has a higher octane rating and a lower volatility. It is also less hydrophilic than either MTBE or ethanol and, therefore, less likely to permeate and pollute groundwater supplies. Volatile organic compound (VOC) emissions are also lessened by ETBE’s lower volatility compared with MTBE. ETBE has a slightly lower oxygen content than MTBE (and much lower than ethanol) so that larger volumes are required, but its higher cost of production remains its principal disadvantage when compared with either MTBE or ethanol. However, the high cost can generally be partially offset with renewable fuel subsidies, and ETBE production is becoming increasingly viable in the * Author to whom correspondence should be addressed. Telephone: 619-351-3776. Fax: 619-351-3554. Email: [email protected]. † Curtin University of Technology. ‡ University of Iowa. S0888-5885(96)00283-7 CCC: $14.00

current market. Table 1 (Furzer, 1994; Brockwell et al., 1991; Lide, 1994) summarizes the key differences between ETBE and its main alternatives. A more complete comparison of the physical properties of ETBE and MTBE has been published previously (Sneesby et al., 1995). Process schemes based on reactive distillation (also called catalytic distillation where a catalyst is present) are now acknowledged to offer a technological advantage for MTBE production compared with conventional synthesis routes (Zhang et al., 1995) as yields are higher and operating costs are lower. The high degree of internal recycle created by the distillation operation helps overcome the thermodynamic restriction of a relatively low equilibrium constant and allows the reaction to proceed further than would otherwise be possible. The reactive distillation process is also more energy efficient than a conventional system, as heat generated by the exothermic reaction offsets the reboiler heat input and contributes to product separation. High conversion and a high ether product purity can be obtained simultaneously in a single device. The extension of reactive distillation technology to ETBE synthesis appears to be a natural next progression. Simulation results have been published for various reactive distillation columns including several systems for MTBE synthesis (Abufares and Douglas, 1995; Sundmacher and Hoffman, 1995). However, little has been published on ETBE synthesis. As the combination of reaction and distillation in a single vessel can produce interactions between various design variables which lead to unusual responses to changes in operating conditions, it is important to fully understand these responses to avoid suboptimal performance and poor designs. Furthermore, differences in phase behavior between the MTBE and ETBE systems lead to different sets of operating conditions and considerations, and the simple extrapolation of concepts from MTBE synthesis to ETBE synthesis may prove to be misleading (Sneesby et al., 1995). Some of these issues have been addressed © 1997 American Chemical Society

1856 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997

here via simulation studies using both Pro/II (Simulation Sciences, 1994) and SpeedUp (Aspen Technology, 1993). Reaction Chemistry ETBE is produced from the reversible reaction of isobutylene and ethanol over an acid catalyst, such as the acidic ion-exchange resin, Amberlyst 15:

(CH3)2CdCH2 + C2H5OH S (CH3)3COC2H5(1)

(1)

The reaction is equilibrium limited in the industrially significant range of temperatures so that the equilibrium conversion from a stoichiometric mixture of reactants at 70 °C is only 84.7%. The reaction has been studied, and detailed expressions are available for the equilibrium constant (eq 2: Jensen and Datta, 1995) and the reaction kinetics based on the LangmuirHinshelwood-Hougen-Watson (LHHW) model (Jensen and Datta, 1996). Two reaction rate expressions are givensthe generalized case (eq 3) and a simplified case (eq 4) for conditions where the ethanol mole fraction is greater than 4%. The proposed reaction mechanism involves two adsorbed ethanol sites reacting with one adsorbed isobutylene in the rate-determining step, giving a total of three active sites. The rate equation derived from this model produced a better fit with the experimental data than other models.

Equilibrium constant 4060.59 T 2.89055 ln T - 0.0191544T + 5.28586 × 10-5T2 - 5.32977 × 10-8T3 (2)

Table 1. Properties of Alternative Oxygenates property

ETBE

MTBE

ethanol

molecular weight oxygen content (wt %) normal boiling point (°C) blending RVP (kPa) octane ((RON + MON)/2) energy content (MJ/kg) relative cost renewable source?

102 15.7 73 27 111 36.3 moderate partly

88 18.2 55 55 110 35.3 low no

46 34.8 78 122 115 26.7 moderate yes

Table 2. Key Physical Properties of the Reaction Components property

ethanol

isobutylene

ETBE

molecular weight specific gravity normal boiling point (°C) boiling point at 1000 kPa (°C) blending RVP (kPa) specific heat (kJ/kg) octane ((RON + MON)/2) Energy content (MJ/kg)

46 0.795 78 155 122 2.46 115 26.7

56 0.600 -7 75 440 1.27 n/a 44.6

102 0.746 73 174 27 2.10 111 36.3

literature. An expression for the equilibrium constant of the dimerization has previously been estimated from the free energies of formation (eq 6: Columbo et al., 1983).

ln K ) 95.2633 + 5819.8644/T 17.2 ln T - 0.0356T (6) If any water is present in the reaction environment, one further, undesirable reaction (the hydration of isobutylene to form isobutyl alcohol) is possible:

KETBE ) 10.387 +

Generalized rate equation

(

mcatkrateaEtOH2 aiBut rETBE )

aETBE KETBEaEtOH

)

(1 + KAaEtOH)3 ln KA ) -1.0707 +

1323.1 T

(

krate ) 7.418 × 1012 exp -

(3a)

(3b)

60.4 RT

)

(3c)

Simplified rate equation (for EtOH > 4 mol %)

(

rETBE ) mcatkrate

)

aiBut aETBE 2 aEtOH K ETBEaEtOH

(

krate ) 1.209 × 1012 exp -

87.2 RT

)

(4a)

(4b)

The ETBE (and MTBE) reaction system also includes an unavoidable side reactionsthe dimerization of isobutylene to produce diisobutylene (DIB):

(CH3)2CdCH2 + (CH3)2CdCH2 S [(CH3)2CdCH2]2 (5) This reaction is also equilibrium limited, but its kinetics and mechanism are not readily available in the open

(CH3)2CdCH2 + H2O S (CH3)3COH

(7)

It can be seen from both ETBE rate equations (eqs 3 and 4) that ethanol has a retarding effect on the reaction. However, in practice, some ethanol excess is required to prevent significant side reactions involving isobutylene. The LHHW reaction model predicts a large adsorption equilibrium constant for ethanol, which implies that, at ethanol excesses of 4 mol % and above, the catalyst surface is largely covered with ethanol. Under these conditions, the dimerization and oligomerization of isobutylene are essentially eliminated (Kitchaiya and Datta, 1996). The ETBE reaction rate also depends on the activity of the catalyst which is susceptible to both deactivation (slow aging) and poisoning (fast aging). Poisoning is potentially a serious problem as water and, especially salts, neutralize active catalyst sites. Deactivation occurs over a much longer period and is accelerated by thermal degradation caused by hot spots due to inadequate mixing. In situ regeneration has generally been unsuccessful and a regular catalyst changeover is required for most reactors (Flato and Hoffman, 1992). The key physical properties of the reaction components differ significantly and are summarized in Table 2 (Simulation Sciences, 1994; Furzer, 1994; Brockwell et al., 1991; CRC Handbook, 1995). In an industrial situation, isobutylene is likely to only be available as a component within a mixture of other, nonreactive butylenes. For most practical purposes, the physical properties of other butylenes present can be assumed to be similar to those of isobutylene. There is variability in the octane data reported in the literature for each of the various ethers, but ETBE is generally claimed to

Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1857 Table 3. Significant Azeotropes in the ETBE Reaction System composition at 0 kPa EtOH-iBut EtOH-nBut EtOH-ETBE EtOH-iBut EtOH-nBut EtOH-ETBE

composition at 950 kPa

composition at 1400 kPa

Reported from Experimental Data n/a n/a 0.94% EtOH n/a n/a n/a 37% EtOH n/a n/a Predicted by UNIFAC no azeotrope no azeotrope no azeotrope no azeotrope 38% EtOH 59% EtOH

1.25% EtOH 1.45% EtOH 66% EtOH

have an average octane rating one number higher than MTBE (Furzer, 1994; Brockwell et al., 1991; Piel and Thomas, 1990; Unzelman, 1989). An octane rating is not generally cited for isobutylene due to its high volatility, but a blending value of around 100 is anticipated based on isoolefin trends. The ETBE system is susceptible to azeotropes due to nonidealities in the liquid phase. Azeotropes between ethanol and isobutylene and between ethanol and ETBE have been recorded experimentally (Gmehling et al., 1994). The UNIFAC model predicts the presence of these azeotropes and also suggests an azeotrope between ethanol and 1-butylene. The literature data are incomplete, but the UNIFAC model indicates that the azeotropes between ethanol and butylenes only exist at high pressure. The compositions of all azeotropes vary with pressure according to the data shown in Table 3.

ETBE Synthesis Route Using Reactive Distillation The conventional process for synthesis of MTBE consists of four stages: (a) pretreatment; (b) reaction; (c) purification; (d) recovery of unreacted reactants (ARCO Chemical Technology, 1995). A simplified process flow diagram is shown in Figure 1. The hydrocarbon feed should be rich in isobutylene. In practice, this implies a refinery sourcesvarious fluidized catalytic cracking (FCC) units (15-35% isobutylene, depending on the type of catalyst in use), a steam cracking unit (40-55% isobutylene), or an isobutane dehydrogenation unit (40-55% isobutylene) (Miracca et al., 1996). The ethanol feed should be essentially pure to minimize the side reaction of isobutylene hydration. The first stage of the conventional process, pretreatment, consists of a simple water wash and is necessary to remove trace contaminants that could deactivate the catalyst. The second stage, reaction, is conducted in two sequential reactors to ensure that high conversions are achieved. The first reactor normally operates isothermally and is used to perform the majority of the reaction. A tubular reactor is normally used to allow the substantial heat which is liberated to be removed adequately before it affects the reaction equilibrium. An adiabatic reactor can be used, if a slipstream of the product is cooled and recycled to the reactor inlet to control the temperature rise, but a lower conversion is expected and the risk of hot spots deactivating the catalyst is increased (Sarathy and Suffridge, 1993). The second stage can mostly be operated adiabatically as much less heat is liberated and a packed-bed reactor is more economical. The packed bed allows more catalyst to be used so that the reactor can be operated at lower temperatures to improve the reaction equilibrium and

maximize conversion. The first reactor operates up to 90 °C, while the second reactor operates at 50-60 °C. The product from the second reactor is then purified by distillation in the third stage of the process. The bottoms product from the ether purification column is high-purity MTBE (>99.5%). The distillate product is further processed to recover unreacted methanol in the final stage. A liquid-liquid extraction column with water as the solvent is used due to the presence of an azeotrope between methanol and butylenes (at around 10% methanol). The raffinate product from this column contains essentially all the nonreactive hydrocarbon from the original hydrocarbon feed and is suitable for alkylation as the concentration of isoolefins is very low. One further column is required to separate water, which is recycled to the L-L extractor, from the methanol, which is recycled to the reactors (ARCO Chemical Technology, 1995). The reactive distillation process for MTBE synthesis combines elements of the second and third stages (reaction and purification) into one element of process equipment, according to the simplified flow diagram shown in Figure 2 (ARCO Chemical Technology, 1995). The second reaction stage and peripheral equipment is eliminated, and the overall conversion is actually increased as the effective methanol excess is much higher within a distillation system. It is economical to maintain the first reaction stage as the amount of catalyst that can be loaded into a reactive distillation column is limited and the operating temperatures therein are restricted by the intersection with separation conditions. The recovery equipment downstream of the reactive distillation column operates essentially unchanged, and product purities are as high as would be achieved with a conventional distillation operation. Conventional ETBE synthesis follows a similar route to MTBE synthesis and uses the same process equipment. However, reaction and phase differences create slightly different operating conditions and change the expected conversion and purity. The more restrictive equilibrium lowers the conversion achieved in the reaction stage by 1-2%. The absence of any significant azeotrope between ethanol and butylenes results in unreacted ethanol being recovered with the ETBE product. The presence of an azeotrope between ethanol and water results in some water being recycled to the reaction stage via extract from the liquid-liquid extractor. However, there is a much lower load on the recovery equipment, as only a small amount of ethanol appears with the distillate product from the main purification column. In fact, it is feasible, under some refinery conditions, to eliminate the ethanol recovery equipment altogether, creating the simplified flow diagram shown in Figure 3. A reactive distillation process for ETBE synthesis would utilize the same principles as the MTBE process and should yield the same benefits of increased conversion, increased energy efficiency, and reduced capital cost. The majority of the reaction (say, 80%) would be performed in an isothermal, tubular reactor operating at moderate conditions of temperature (around 90 °C) and pressure (1500-2000 kPa to ensure all components remain in the liquid phase). The feed to the reactive distillation column would, therefore, be rich in ETBE but still contain some ethanol and isobutylene. The products from the bottoms and overheads of the reactive distillation column would be ETBE with some ethanol and nonreactive hydrocarbon with a small amount of


Figure 1. Conventional MTBE synthesis route.

Figure 2. MTBE synthesis via reactive distillation.

Figure 3. Conventional ETBE synthesis route without ethanol recycle.

isobutylene and ethanol, respectively. The distillate product may or may not require further processing depending on its composition and the refinery configuration. This scheme is shown in Figure 4 without ethanol recovery equipment. Reactive Distillation Model Both the MESH (Material balance, vapor-liquid Equilibria equations, mole fraction Summations, and

Heat balance) equations for systems in vapor-liquid equilibrium and the MERQ (Material balance, Energy balance, Rate equations for mass transfer, and phase eQuilibrium at the vapor-liquid interface) equations for mass transfer can be used to model distillation processes. The MERQ equations have been recommended by some researchers (Sundmacher and Hoffman, 1995) and are gaining in popularity due to advances in computational power available. However, they are more complex, require the estimation of more empirical

Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1859

parameters, and appear to offer no improvement in accuracy for most systems. The MESH model was used here (see eqs 8-37) and produced satisfactory results.

Total Condenser

Separation Stage

component balance Lin + Vin - Lout - Vout ) 0

material balance component balance

energy balance

(8)

Linxi,in + Vinxi,in Loutxi,out - Voutxi,out ) 0 (9)

L V LoutHi,out - VoutHi,out ) 0 (10)

Pyi ) γixiPvap i

(11)

TV ) TL

(12)

∑yi ) 1

(13)

(28)

Vyi - Lxi ) 0

(29)

VHV - LHL - Qc ) 0

(30)

equilibrium

P)

material balance

∑γiyiPvap i

Lin - Lout - Vt ) 0

component balance

energy balance

(31)

(32)

Linxi,in - Loutxi,out - Vyi ) 0 (33)

L LinHin - LoutHLout - VHV + Qr ) 0 (34)

Pyi ) γixiPvap i

(35)

TV ) TL

(36)

(14)

∑yi ) 1

(37)

r1,ETBE ) -r1,EtOH ) -r1,iBut

(15)

r2,DIB ) -2r2,iBut

(16)

The basic set of MESH equations was modified for reactive distillation by adding equations to model the chemical reactions present. The equations shown here are fully rigorous in that both reaction equilibrium and reaction rate expressions are included, but chemical equilibrium could have been assumed for this system due to the relative speed of the reaction. This would have been done by eliminating eqs 18, 20, and 21 (effectively assuming an infinite supply of catalyst). The construction of the equations is such that the heat of reaction is calculated implicitly and does not need to be separately specified. The overhead pressure and stage-to-stage pressure drop were fixed for these simulations but could have been modeled more rigorously by incorporating variable relationships between holdup and vapor-liquid flows (see part 2 of this series). However, constant stage-to-stage pressure drop is a common assumption and does not introduce any significant error for steady-state simulations. The choice of physical property routines is important due to the highly nonideal nature of the ether-alcoholalkene system. The UNIFAC model has been used successfully to predict liquid phase activities by several researchers including developers of reaction equilibrium constant expressions and rate equations for the ETBE (Jensen and Datta, 1995, 1996) and MTBE (Zhang and Datta, 1995) reactions. It has also been found (see Table 3) to accurately represent the azeotropes compared with experimental data (Gmehling et al., 1994). Alternatively, the UNIQUAC model (Bravo et al., 1993) or the Wilson equation (Abufares and Douglas, 1995) can be used to predict liquid phase activities in ether systems. Vapor phase nonideality is less significant as etherification columns operate at only modest pressures and any one of a range of prediction methods should be adequate. The Soave-Redlich-Kwong (SRK) method is widely used for the prediction of enthalpy and other properties although alternative methods (e.g., PengRobinson) would also be acceptable for the pressure and temperature conditions.

equilibrium

Reactive Stage Lin + Vin - Lout - Vout +

material balance

∑r1,j + ∑r2,j ) 0

r1,nBut ) r1,DIB ) r2,ETBE ) r2,EtOH ) r2,nBut ) 0

(

mcatkrateaEtOH2 aiBut r1,ETBE )

aETBE KETBEaEtOH

(1 + KAaEtOH)3

)

(17)

(18)

4060.59 T 2.89055 ln T -0.0191544T + 5.28586 × 10-5T2 - 5.32977 × 10-8T3 (19)

ln KETBE ) 10.387 +

ln KA ) -1.0707 +

1323.1 T

(

krate ) 7.418 × 1012 exp ln KDIB ) 95.2633 +

component balance

energy balance

(20)

60.4 RT

)

(21)

5819.26 - 17.2 ln T - 0.0356T T (22) Linxi,in + Vinxi,in - Loutxi,out Voutxi,out ) r1,j + r2,j ) 0 (23)

L V + VinHi,in LinHi,in L V - VoutHi,out ) 0 (24) LoutHi,out

equilibrium

energy balance

V-L)0

Partial Reboiler

L V LinHi,in + VinHi,in -

equilibrium

material balance

Pyi ) γixiPvap i

(25)

TV ) TL

(26)

∑yi ) 1

(27)


Figure 4. ETBE synthesis via reactive distillation.

The results presented below for various ETBE reactive distillation columns were obtained from two different commercial simulators: Pro/II (Simulation Sciences, 1994) and SpeedUp (Aspen Technology, 1993). The Pro/ II simulation results were obtained using the built-in reactive distillation unit operation and required only the specification of components, reaction equilibria, feed conditions, operating pressure, column configuration, and two operating parameters (from reboiler and condenser duties, reflux rate/ratio, and various product specifications) to generate simulation results. The UNIFAC model was used for liquid phase activities. The SRK method was used for fugacity coefficients, enthalpy, and other properties. The reaction expressions given by eqs 2 and 6 were used to model the reactions present. The SpeedUp model was developed from scratch using eqs 8-37. The UNIFAC model was again used to predict liquid phase activities. However, an ideal vapor phase was assumed to simplify the model. Published Antoine coefficients (Krahenbu¨hl and Gmehling, 1994; Gmehling and Onken, 1977; Dean, 1992; Reid et al., 1987) were used to predict vapor pressures and, therefore, vapor-liquid equilibrium. Rigorous reaction kinetics were used to model the ETBE reaction (eqs 1114), while reaction equilibrium was assumed for the DIB side reaction (eq 15). The global system of equations for the full model contained a total of 578 variables and 504 linear and nonlinear equations. An ETBE reaction equilibrium model, requiring six fewer equations, was also built to test the assumption of chemical equilibrium used for the Pro/II simulations. The isobutylene conversion was found to be only 0.2-0.3% lower for the second case using a moderate catalyst loading. In comparing the two process simulators, the SpeedUp model offers a greater flexibility and the potential to expand the model to investigate dynamic responses, but it required significantly more time to build and develop these models and greater prior knowledge of the system compared with Pro/II. Simulation Results The laboratory column originally used to patent the catalytic distillation process for MTBE synthesis (Smith, 1980) was used as the basis for the ETBE simulations presented here. Smith’s column contained approximately 10 ideal stages, one of which was packed with catalyst (Abufares and Douglas, 1995). To fit with the

synthesis route shown in Figure 4, the feed is to stage 6 and consists of a 80% prereacted mixture of isobutylene and 1-butylene (40/60 split) plus ethanol at a stoichiometric excess of 5.0%. Three reactive stages were specified for the ETBE column so that a higher loading of catalyst was possible to accommodate the lower reaction rate of ETBE compared with MTBE. This column is shown in Figure 5. The feed rate and reboiler duty were based on a pilot scale (154 mm diameter) ETBE reactive distillation column which has recently been constructed at Curtin University, Western Australia. Operating conditions were adjusted slightly from those proposed for MTBE synthesis to compensate for differences between the two systems and improve the overall viability of the process. A slightly higher pressure (950 kPa compared with 690 kPa in Smith’s column) was used to increase the reaction zone temperature and, thereby, improve the reaction rate and reduce the catalyst requirement. A lower reflux ratio (5.0 compared with 10.0) was also specified to reduce the energy requirements. The reboiler duty was optimized with respect to isobutylene conversion. The complete simulation input is shown in Table 4. The key simulation results for the ETBE column, from both Pro/II and SpeedUp, are shown in Table 5. The estimates of conversion and purity from the two simulators are both within 1%, and the key indicator of bottoms temperature is only 1 °C divergent. The temperature profiles and composition profiles from each simulation are compared in Figures 6 and 7, respectively, and also match well. The most significant discrepancies are between estimates of ethanol concentration around the middle of the stripping section and the temperature profile near the top of the column. The variation in predicted ethanol concentration in the stripping section can be attributed to the assumption of an ideal vapor phase in the SpeedUp model as this assumption was tested using Pro/II. Vapor phase nonideality is low where the ethanol concentration is less than 10 mol % so that the composition and temperature profiles are generally acceptable over the rest of the column. The differences in temperature profiles are likely to be the result of the slightly different vapor pressure correlations that were used (Pro/II used builtin data, while published Antoine coefficients were used in the SpeedUp model). The lower isobutylene conversion predicted by the SpeedUp model is at least partially attributable to the incorporation of the full reaction kinetics in that model as a satisfactory solution to the


Figure 5. Reactive distillation column configuration for ETBE synthesis. Table 4. ETBE Reactive Distillation Column Simulation Input feed conditions

column specification

temperature (°C) rate (L/min) composition (mole basis)

30 0.76 29.1% ETBE, 9.1% ethanol, 7.3% isobutylene, 54.5% n-butylenes

composition (weight basis)

43.3% ETBE, 6.1% ethanol, 6.0% isobutylene, 44.6% n-butylenes

overall excess ethanol (mol %)

5.0

rectification stages reaction stages stripping stages total stages overhead pressure (kPa) condenser reflux ratio reboiler reboiler duty (kW)

2 3 5 10 950 total 5.0 partial 8.26

Table 5. ETBE Reactive Distillation Column Simulation Results property

Pro/II

SpeedUp

condenser temp (°C) reaction zone temp (°C) reboiler temp (°C) isobutylene conversion (mol %) bottoms composition (wt %)

74 75-81 159 98.3 96.1% ETBE, 2.1% ethanol, 1.7% butylenes, 0.06% DIB 97.9% n-butylenes, 0.9% isobutylene, 1.2% ethanol 6.86 8.26 0.16 (stage 3), 0.19 (stage 4), 0.15 (stage 5)

79 80-84 160 97.4 96.5% ETBE, 2.8% ethanol, 0.7% butylenes, 0.04% DIB 97.6% n-butylenes, 1.7% isobutylene, 0.7% ethanol 6.73 8.33 0.17 (stage 3), 0.18 (stage 4), 0.12 (stage 5)

distillate composition (wt %) condenser duty (kW) reboiler duty (kW) reaction rates (mol/min)

full kinetic case could not be produced using Pro/II due to difficulties in obtaining convergence. The overall agreement between the two simulators is excellent and is considered to be an indication of model validity. The key blending properties of the reactive column bottoms product can be compared with pure ETBE and pure MTBE. These properties were estimated using the Pro/II database and are shown in Table 6 (Simulation Sciences, 1994). The actual values quoted for pure ETBE and MTBE do not necessarily correspond to experimental data but are produced on the same basis as the column bottoms’ properties for comparative purposes only. The total fuel concentration for the ETBE column bottoms product is defined as the ethanol concentration plus ETBE concentration. The data in

Table 6 indicates that the presence of some ethanol in the column bottoms product has a negligible impact on the overall properties but the small percentage of butylenes (1.8 wt %) significantly reduces the initial boiling point and flash point and increases the RVP of the mixture. This is a potential disadvantage and may necessitate modification of the column operation and/ or design to increase the ETBE product purity. Such modifications would include additional separation stages and may have the unfortunate side effect of reducing the isobutylene conversion. Model Validation and Comparison To validate the simulation results without experimental data, Smith’s MTBE column was simulated for

1862 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 Table 6. Gasoline Blending Properties of the ETBE Column Product

Figure 6. Temperature profile in an ETBE column.

Figure 7. Composition profiles in an ETBE column.

the case described in his patent application (Smith, 1980) using both Pro/II and SpeedUp. The numbers of rectification, reaction, and stripping stages were based on previous estimates (Abufares and Douglas, 1995). The equation structure and physical property routines used were identical to those used for the ETBE column simulations described above. Table 7 shows both sets of simulation results for Smith’s column compared with the limited experimental data from the patent application. The agreement between Pro/II, SpeedUp, and the experimental results is generally excellent, especially for the key indicators of bottoms temperature, isobutylene conversion, and MTBE purity. The most significant discrepancy between the experimental data and the simulation models is the estimate of DIB concentration in the bottoms product. However, both sets of simulation results presented here predict DIB concentrations of around 1%, which is also in agreement with the previously published simulation results, suggesting an alternative explanation for the values reported for the experimental system. The acceptable accuracy of the MTBE simulations demonstrates the adequacy of the modeling process and validates its extension to ETBE reactive columns where the column configuration, operating conditions, and phase behavior are similar. Subsequently, there is a high degree of confidence in the ETBE simulation results presented here. However, a pilot plant has been built, and the results from planned experiments will permit the simulations to be checked even more rigorously. The present study also provides a basis for comparing the simulation packages. Pro/II is a sequential-modelbased simulation package that contains a wide range of built-in models which allows most unit operations to be modeled easily and accurately. The time to develop and run simulations is generally short, and the models

property

ETBE column bottoms

pure ETBE

pure MTBE

initial boiling point (°C) final boiling point (°C) RVP (kPa) flash point (°C) total fuel concentration (wt %) energy content (MJ/kg) oxygen content (wt %) octane rating ((RON + MON)/2) specific gravity

-6 80 41 -48 98.2 36.2 15.8 111 0.743

73 73 29 -19 100 36.3 15.7 111 0.746

55 55 55 -30 100 35.3 18.2 110 0.747

Table 7. Key Simulation Results and Experimental Data for Smith’s MTBE Column property

Pro/II

SpeedUp

experimental

condenser temp (°C) reaction zone temp (°C) reboiler temp (°C) isobutylene conversion (mol %) bottoms MTBE purity (mol %) DIB bottoms concentration (mol %) condenser duty (kW) reboiler duty (kW)

61.0 69.0 128.0 91.6 92.2 1.2 2.14 2.32

64.6 70.1 125.7 91.6 92.6 0.9 1.99 2.16

n/a 71 127 91 91.9 6.1 n/a n/a

are robust enough to converge when only a few initial estimates are provided. A reactive distillation module is present, based on a standard distillation module. Either chemical equilibrium or kinetics can be assumed, although the likelihood of convergence is reduced where reaction kinetics are specified. SpeedUp is an equationbased simulator and offers the benefits of greater flexibility, but model development is significantly longer unless library models are used, in which case the flexibility benefits are partially lost. Convergence criteria is much more stringent, and many more preset variables are needed to produce a converged solution. However, SpeedUp models can be expanded to allow the prediction of dynamic responses and can handle reaction kinetics with greater ease provided the equation structure is correct. For the majority of steady-state simulation objectives, Pro/II is the preferred simulator, but SpeedUp offers potential for the study of process control applications, startup analysis, and hysteresis phenomena. Homotopy Analysis Reactive distillation columns behave substantially differently from conventional distillation columns due to the interactions between the chemical reactions present and vapor-liquid equilibrium. The effects of key design and operating variables are discussed in detail below with reference to the ETBE column studied above. There are also substantial differences between ETBE reactive distillation and the more common MTBE reactive distillation, and comparisons are made where relevant. The effects described below should be considered during design and operation of the column to ensure optimal performance. Effects of Feed Composition. The hydrocarbon feed composition to an etherification unit is essentially fixed by upstream plant operations and usually varies between 15% and 55% reactive isobutylene depending on the type of unit and type of catalyst in use upstream. Standard FCC units produce C4 streams with 15-20% isobutylene, while FCC units equipped with new generation catalysts produce C4 streams with up to 35%

Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1863 Table 8. Effect of Hydrocarbon Feed Composition on the Maximum Conversion and Corresponding Ether Purity in Two ETBE Columns 10-stage column

30-stage column

isobutylene max ether max ether concentration conversion product conversion product (mol %) (mol %) purity (wt %) (mol %) purity (wt %) 15 20 30 40 50 60

98.5 98.8 98.7 98.3 97.5 95.7

95.0 96.2 96.7 96.1 95.4 94.1

98.7 98.6 94.5 91.7 88.5 86.6

97.2 96.9 94.8 93.6 91.6 91.1

isobutylene. The C4 streams from steam cracking and isobutane dehydrogenation units are even richer in isobutylene (up to 55%) (Miracca et al., 1996). In each case, the other components are predominantly n-butylenes although other C4s (mainly isobutane and nbutane) are also present in some cases. Although the operation of the various catalytic and steam cracking units can be varied to increase or decrease isobutylene production, other factors (e.g., maximizing gasoline production) usually have a more significant economic impact and govern the operation of the unit. Increasing the concentration of reactive isobutylene in the hydrocarbon feed has four main effects on reactive distillation column operation: (a) the energy cost, per kg of ETBE produced, decreases as less energy is “wasted” heating and cooling inert components; (b) the reactant concentrations in the reaction zone increase, with a favorable effect on the reaction equilibrium; (c) the reaction zone temperatures and the reaction zone temperature gradient increase as the stabilizing effect of inert components is lessened, with a detrimental effect on the reaction equilibrium; (d) the specific reboiler duty must be decreased to maintain optimal reaction conditions, with a detrimental effect on product purity. The overall effect of increasing isobutylene in the feed is usually to decrease the maximum isobutylene conversion and corresponding product purity. The effect of isobutylene concentration on reactive distillation column operation contrasts with MTBE synthesis where an isobutylene feed concentration of around 60% is optimal for conversion and energy efficiency. A consequence of this result is that, when the isobutylene concentration in the hydrocarbon feed is low, ETBE synthesis may be more favorable than MTBE synthesis for some column configurations. However, the presence of a significant azeotrope in the MTBE system, between methanol and butylenes, means that isobutylene conversion and MTBE product purity can essentially be increased as far as is desired by adding more stages and/or increasing reflux. This does not necessarily apply to ETBE systems (see discussion in the Effects of Separation Stages section). The maximum conversion in a 10-stage ETBE reactive distillation column (Figure 5) and a 30-stage ETBE reactive distillation column based on a commercial MTBE column (Simulation Sciences, 1995) (where the co-objective is to essentially eliminate butylenes from the ether product) was determined for varying isobutylene concentrations in the hydrocarbon feed to the primary reactor, by simulations using Pro/II. For all cases, 80% conversion in the reactor was assumed and the reactive distillation reboiler duty was optimized with respect to the final, overall conversion. Table 8 shows the maximum final conversions and the corresponding ETBE product purities.

ETBE synthesis from pure isobutylene and pure ethanol (three-component system only) is not feasible, as the intersection of phase and chemical equilibrium is not favorable. Too little isobutylene is available in the liquid phase for reaction, and, at pressures that result in favorable reaction equilibrium, the separation of ethanol and ETBE is difficult. The maximum isobutylene conversion and ether purity attainable in a ternary system from the column configuration shown in Figure 5 are around 83 mol % and 91 wt %, respectively. Effects of Stoichiometric Excess of Ethanol. The stoichiometric ratio of reactants significantly affects the reaction conversion and the loads on the downstream product purification and reactant recovery equipment. If the reactant excess is too low, product conversion is adversely limited, while if it is too high, purification costs are increased and/or product purity is decreased. The choice of percent excess essentially becomes a compromise between operating costs and the value added by the process (a function of market conditions). The control of the stoichiometric ratio is often complicated by the inability to accurately measure feed composition and the need to ensure that certain constraints (resulting from reaction kinetics or other sources) are never violated. In an MTBE column, the majority of unreacted methanol is recovered in the distillate product via azeotropes that occur between methanol and butylenes. This places an upper limit on the methanol excess as sufficient butylenes must always be present in the distillate product to maintain the unreacted methanol below the azeotropic composition (around 10 mol % methanol). For example, if the hydrocarbon feed contained 60% isobutylene and 40% n-butylenes and the azeotropic composition at the operating pressure was 10 mol % methanol, then the maximum methanol excess would be approximately given by:

unreacted MeOH e 10% × (nBut + unreacted iBut) e 10% × (≈41% total hydrocarbon feed) e 4.1% × (total hydrocarbon feed) reacted MeOH ) reacted iBut ≈ 59% × (total hydrocarbon feed) maximum MeOH excess ≈ (4.1%/59%) ) 6.9% Another consequence of recovering methanol overhead via azeotropes is that, below the maximum methanol excess as determined by the feed and azeotropic compositions, increasing the methanol excess has only a limited effect on the bottoms product purity. However, in an ETBE column, unreacted ethanol is recovered directly with the ether product. This removes the restriction on ethanol excesses that can be used but increases the effect that the ethanol excess has on the ether purity. Therefore, a compromise must be determined between isobutylene conversion (which rises as the ethanol excess increases) and ether purity (which falls as the ethanol excess increases). The column configuration influences this decision by changing the relative magnitudes of the two effects, but an ethanol excess between 4.0 and 7.0 mol % is considered sufficient to produce high isobutylene conversions without

1864 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 Table 9. Effect of Column Overhead Pressure on the Maximum Isobutylene Conversion and Corresponding Ether Purity in Two ETBE Columns 10 stage-column

30 stage-column

overhead pressure (kPa)

reaction zone temp (°C)

maximum conversion (mol %)

ether purity (wt %)

reaction zone temp (°C)

maximum conversion (mol %)

ether purity (wt %)

400 500 600 700 800 900 1000 1100

44-53 51-59 57-65 63-70 68-74 72-79 77-82 81-86

99.0 99.2 99.1 99.0 98.8 98.5 98.2 97.8

97.0 97.1 97.1 97.0 96.6 96.4 95.9 95.3

44-50 51-56 57-62 62-67 67-72 72-77 77-81 81-85

99.1 98.9 98.7 98.6 98.6 98.2 98.0 97.7

97.1 97.1 97.3 97.6 98.1 98.3 99.1 99.9

Figure 8. Effects of the stoichiometric excess of ethanol on isobutylene conversion. Ether purity and distillate composition from an ETBE column.

Figure 9. Effects of reactive stages on isobutylene conversion from an ETBE column.

overly diluting the ether product with ethanol (thereby adding to downstream recovery costs) while providing a satisfactory driving force for the reaction. A very highpurity ether product can be produced with a lower ethanol excess, but some excess of ethanol should always be used to suppress side reactions involving isobutylene. Reaction zone conditions and distillate composition are essentially independent of the percent excess of ethanol in an ETBE column. Using the column configuration shown in Figure 5 as the basis for simulations, Figure 8 shows the effect of increasing the ethanol excess on isobutylene conversion and ETBE purity. Effects of Column Pressure. In conventional distillation, the operating pressure of a column is normally set through an economic rationalization of heat-transfer costs and the value of improved separation (via increasing relative volatility with reducing pressure) (Kister, 1992). However, in reactive distillation, the choice of operating pressure is complicated by the indirect effect of pressure on the reaction equilibrium via changing phase equilibrium temperaturessincreasing the pressure raises the reaction zone temperature and

Figure 10. Effects of rectification separation on isobutylene conversion and distillate composition from an ETBE column.

Figure 11. Effects of stripping separation on isobutylene conversion and ether purity from an ETBE column.

decreases the reaction equilibrium constant of exothermic reactions such as ETBE synthesis, thereby lowering conversion. The effect of pressure on the rate constant, via VLE temperature changes, must also be considered, if the reaction is kinetically controlled. For both ETBE and MTBE systems, the nonideality of the liquid phase adds two further restrictions to column operations. First, below a certain pressure (about 130 kPa for ETBE and 300 kPa for MTBE columns) the highest boiling component is the alcohol rather than the ether. Operating below this pressure will drive the ether product back to the reaction zone and will result in low conversion and low ether purity. Second, the overhead pressure affects the composition (and presence) of the alcohol-butylene azeotropes. This is of greater significance for MTBE columns as the azeotrope is more common and contains a larger fraction of unreacted alcohol but should not be ignored for ETBE columns. The range of effects that need to be considered in selecting the column operating pressure suggests that accurate simulation results are almost essential for reactive distillation column design. Table 9 shows the

Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1865 Table 10. Effect of Separation Stages on the Maximum Conversion and Corresponding Ether Purity in ETBE Columns with Lean Isobutylene Feeds rectification stages 2 4 8

4

stripping stages 8

16

98.6 mol % conversion; 90.3 wt % purity (8.8 wt % butylenes) 98.8 mol % conversion; 92.2 wt % purity (6.8 wt % butylenes) 99.0 mol % conversion; 98.6 wt % purity (0.16 wt % butylenes)


99.4 mol % conversion; 99.95 wt % purity (∼0 wt % butylenes) 99.5 mol % conversion; 99.7 wt % purity (∼0 wt % butylenes) 99.5 mol % conversion; 98.9 wt % purity (∼0 wt % butylenes)

Table 11. Effect of Separation Stages on the Maximum Conversion and Corresponding Ether Purity in ETBE Columns with Rich Isobutylene Feeds rectification stages 2 4 8

4

stripping stages 8

16



98.2 mol % conversion; 97.2 wt % purity (∼0 wt % butylenes) 98.2 mol % conversion; 97.1 wt % purity (∼0 wt % butylenes) 98.2 mol % conversion; 97.1 wt % purity (∼0 wt % butylenes)

maximum conversion attainable in two ETBE columns for varying overhead pressures, assuming chemical equilibrium is attained. The first column is the base case shown in Figure 5, while the second column corresponds to the first row of Table 8 (30 stages and 15% isobutylene in the feed). The optimum with respect to conversion is 400-500 kPa, although higher pressures are preferred for kinetically controlled reactions as the reaction rate at temperatures corresponding to this pressure are very low. The optimum with respect to ETBE purity occurs at a significantly higher pressure in the 30-stage column, and very high purities (greater than 99.9 wt %) are possible at high pressures (greater than 1100 kPa). This suggests that an intermediate pressure (e.g., 700-900 kPa) is favored overall. The range of effects present prevents the recommendation of a standard operating pressure for ETBE columns, although an optimum will usually exist. Site factors and the expected variation in feed composition are likely to be decisive in optimizing the operating pressure. However, as a rule-of-thumb, it is considered that operating at a column pressure such that the reaction zone temperatures are 5-15 °C lower than those used in the prereactor should guarantee high conversion, high ether product purity, and manageable reaction rates. Effects of Reactive Stages. The function of the reactive section of the column is simply to provide a site for the main reaction to proceed, and, as such, there is no particular requirement for separation. This suggests that only one equilibrium stage of a column needs catalyst present, although the actual physical size of this stage could be quite large to meet catalyst requirements. However, simulations show that higher conversions are possible where more than one equilibrium stage is reactive. Figure 9 shows simulation results for an ETBE column where the number of reactive stages was varied. All other variables, including the number of separation stages, reflux ratio, reboiler duty, and feed conditions, were unchanged. The improved conversion with an increased number of reactive stages results from the benefits of the

Figure 12. Effects of reflux ratio on isobutylene conversion. Ether purity and energy requirement of an ETBE column.

Figure 13. Effects of reboiler duty on isobutylene conversion and ether purity from an ETBE column.

additional separation which is gained. Under most conditions, transferring all the catalyst in a column to a single stage would have a negligible effect on the overall conversion achieved. Note that this is different from the data presented in Figure 9 as it implies an increase in the number of separation stages at the expense of reactive stages. Increasing the number of reactive stages above the optimum (four for this column) produced a detrimental interaction between the phase and chemical equilibrium

1866 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 Table 12. Suggested Reactive Distillation Design Strategy for Ether Oxygenates 1. Define values for the key process objectives of hydrocarbon conversion and ether product purity and determine the composition of the hydrocarbon feed stream. 2. Identify the key components in the rectification and stripping sections and the effects on the composition profiles of increasing rectification and stripping separation for the specific feed composition to be used. 3. Select a value for the stoichiometric excess of alcohol reactant considering the conversion and purity requirements identified above (high for maximum conversion; low for maximum purity). 4. Estimate operating pressure based on reaction equilibrium constant at the boiling point of a mixture that is predominantly hydrocarbon with some ether and alcohol. 5. Perform rigorous simulations with a range of numbers of rectification and stripping stages and a range of reflux ratios. 6. The minimum reflux ratio cannot easily be determined for reactive distillation because of interactions with chemical equilibrium. Select a combination of rectification stages, stripping stages, and reflux ratio based on process objectives, local utility costs, and product values. Note that, for some feed compositions and stoichiometric excesses, the rectification and stripping separation must be optimized to prevent excessive loss of unreacted reactants in both products and that for these cases there exists a limit to the conversion and ether purity achievable. 7. Where feed rates are relatively low, ensure the reflux ratio selected is sufficient to adequately load a column of 1.2 m or greater diameter. 8. Determine the number of reactive stages required using simulations with the selected reflux ratio and number of stages. Add reactive stages until decomposition occurs on the lowest reactive stages and/or there is no incremental benefit to isobutylene conversion. 9. Select a value of stage efficiency based on conventional distillation and determine the number of actual stages required for the rectification and stripping sections. 10. Estimate reboiler and condenser duties from simulations based on the above assumptions. 11. Determine column diameter from simulation data for vapor and liquid loadings and column height from stage efficiency estimates, including appropriate allowances for uncertainties in flooding factor and stage efficiency. 12. If required, adjust the reflux ratio to vary column width and height to produce an acceptable design. Note the impact of reducing reflux ratio on conversion over and above its effect on separation and the need to optimize separation for some cases. 13. Based on the column configuration defined above, optimize stoichiometric excess, pressure, and reboiler duty to maximize hydrocarbon conversion and ether product purity using further simulations.

which led to the decomposition of product on the lower reactive stages. This is to be expected as the extra separation occurring with more stages concentrates the ether in the lower reactive stages and shifts the chemical equilibrium back to the reactants. An excessive number of reactive stages can also encourage unwanted side reactions and increase the concentration of impurities in the ether product. By comparison, the MTBE case showed no optimum although the benefits of adding a reactive stage became progressively smaller when high numbers of reactive stages were already present, due to the increased likelihood of decomposition on the lower reactive stages. Supplying catalyst on several reactive stages allows the total catalyst loading in the column to be increased and may, therefore, extend the time between catalyst changeovers/regenerations. However, during the life of the catalyst the main reaction site may shift within the column and change the effective number of rectification and stripping stages and, subsequently, change the conversion and purity attained. Effects of Separation Stages. Ideally, the rectification section of a reactive distillation column for ether synthesis should (a) remove light inerts from the reaction zone, (b) prevent loss of ether or alcohol in the distillate, and (c) recycle unreacted reactants (olefin and alcohol) to the reaction zone. For an ETBE column, this would ideally require a separation between isobutylene and the heaviest nonreactive hydrocarbon lighter than isobutylene. In practice, this is almost impossible to achieve while maintaining acceptable reaction zone conditions. However, the loss of ether in the distillate can be minimized without rejecting isobutylene from the column. More rectification stages are required to also prevent loss of ethanol with the distillate. Ideally, the stripping section should (a) remove ether from the reaction zone to maintain favorable reaction conditions, (b) purify the ether product, (c) prevent loss of reactive olefin with the ether product, and (d) minimize ethanol loss with the ether product. In an ETBE column, this implies a separation between ethanol and ETBE, which is largely achievable at the

conditions of temperature and pressure required for adequate reaction. Manipulating the number of stripping stages provides a mechanism for controlling the volatility, flash point, and composition of the ether product. Although the separation objectives are clear, in a multicomponent system changes in the separation efficiency tend to adjust composition profiles rather than produce more clearly defined product splits when intermediate boiling components (e.g., ethanol) are involved. Consequently, an increase in rectification or stripping separation is not necessarily beneficial for reaction zone conditions. Too much rectification separation can result in isobutylene loss via the distillate. Too much stripping separation can result in ethanol being drawn away from the reaction zone and concentrated just above the reboiler. For some column configurations, both rectification and stripping separation must be optimized. Figures 10 and 11 show examples of this phenomenon for a column based on the configuration given in Figure 5. The ratio between the number of rectification and stripping stages and the feed composition are also important factors that should be considered. If the rectification separation is too great without light inerts present to stabilize the reaction conditions, then reboiler operation must be adjusted to compensate the reaction conditions and, subsequently, the purity of the ether product is sacrificed. Tables 10 and 11 show this effect for two different feed compositions: a low isobutylene feed with 15% isobutylene in the hydrocarbon to the primary reactor and a high isobutylene feed derived from a 50/50 mixture of isobutylene and n-butylenes. The number of reactive stages and the reflux ratio were kept constant for all simulations, and the reboiler duty was optimized with respect to conversion in each case. The best combination of rectification and stripping stages is 2 and 16, respectively. Adding rectification stages is beneficial with few stripping stages present but detrimental with many. With a high concentration of isobutylene in the hydrocarbon feed, the magnitudes


of all effects are diminished and the need to optimize (rather than maximize) separation becomes more likely. Effects of Reflux Ratio. In a reactive distillation column, reflux not only enhances separation but recycles unreacted reactants to the reaction zone and increases conversion. In an ETBE column, even where the concentration of isobutylene is high and rectification separation must be optimized to maximize conversion, increasing reflux still increases conversion. Therefore, separation stages and reflux cannot necessarily be used interchangeably, and high reflux ratios may sometimes be preferred. Several effects occur as a result of increasing the reflux ratio: (a) the concentration of reactants in the distillate is reduced; (b) the reaction zone temperatures are reduced; (c) the concentration of ether in the reaction zone is reduced. Each of these effects contributes to increased conversion. Figure 12 indicates the relationship between reflux ratio and conversion for the reactive distillation column shown in Figure 5. Although conversion increases monotonically with reflux ratio, the ether product purity remains approximately constant over a wide range of reflux ratios due to the increasing load on the stripping section as the amount of ETBE in the column increases. In an industrial environment, a high reflux ratio is economically unattractive as it adds to the equipment size and energy requirements. In conventional distillation, reflux and stages can be traded off against each other. In reactive distillation, this can also be done but may incur a conversion penalty. ETBE columns are susceptible to this penalty, and, depending on the feed composition, the maximum conversion could be achieved in a shorter column with high reflux. No significant conversion penalty exists for MTBE columns, and the most viable design is a tall column with a low reflux ratio. Effects of Reboiler Duty. The reboiler duty is one of the principal points of control in a distillation column. In normal distillation, there is a monotonic, albeit sometimes highly nonlinear, relationship between the reboiler duty and the principal operating objective (Kister, 1992). In reactive distillation, the reboiler duty must be set to ensure sufficient recycle of unreacted, heavy reactant to the reaction zone without excluding the light reactant from the reaction zone. If the reboiler duty is too high or too low, conversion, and subsequently product purity, is reduced. Figure 13 indicates the effect of reboiler duty on conversion and ether purity and clearly shows the presence of an optimum duty. Effects of Other Operating and Design Variables. The extent of reaction performed prior to the reactive distillation column should simply be an economic optimization between the relative costs of the two operations. Initially, it is easy to produce ether in a simple tubular reactor with some form of temperature control to prevent the exothermicity of the reaction from heating the reactor contents and, eventually, stopping the reaction due to equilibrium considerations. However, at higher conversions, very low reaction temperatures are required with a subsequent high demand for catalyst due to the low reaction rates. Under these conditions, reactive distillation becomes more economical. The optimum feed point to the reactive distillation column is just below the reactive section to avoid the possibility of ETBE decomposition with the relatively ETBE-rich product from the reactor. Feeding too far below the reactive section reduces the stripping poten-

Figure 14. Flowchart for reactive distillation design.

tial of the column and increases the energy required for separation. Split feeding to each of the reactive stages is possible but creates a high concentration of ETBE at the top of the reactive section (leading to decomposition toward the bottoms of the reactive section) and, again, increases the energy required for separation. The feed temperature has only a very slight effect on the operation of either an ETBE or MTBE reactive distillation column. A cooler feed has a mildly beneficial effect on the reaction zone temperature, but this can be offset by a shift in phase equilibrium. To minimize equipment requirements, the feed should be supplied at its process temperature, which is likely to be close to the reactor temperature (80-90 °C). If intermediate storage is required for any reason, an ambient feed is also acceptable. Neither feed heaters or feed coolers are required for a satisfactory process design. Process Design Method Simulations should form a key part of reactive distillation process design, more so than in conventional


distillation design, due to the increased complexity of operation and the larger number of design and operating variables. The importance of simulations is further enhanced by the absence of satisfactory shortcut or empirical methods that define the significant effects of key variables. Residue curves and tie-lines (Ung and Doherty, 1995), which parallel conventional distillation design methods, are accepted design tools for the reactive distillation of ternary systems but are only applicable to the reactive section of the column. Little has been published on integrated design methods for reactive distillation that recognize the interactions between all three sections (rectification, reaction, and stripping) of a reactive distillation column. A possible design strategy is proposed in Table 12. It is important to note that the design process is iterative, and a successful design may require several iterations. Figure 14 shows the same design process diagrammatically and indicates where the process may become iterative by showing recycle steps in dashed lines. Conclusions Process simulation of ETBE reactive distillation columns can be performed using either the MESH or MERQ distillation equations, with appropriate additional equations to model the chemical reaction(s). The MESH method was shown to be accurate for a wellknown MTBE column and was extended to ETBE columns using both Pro/II and SpeedUp. The results from the two simulators were in good agreement and were considered to be a satisfactory basis for a homotopic study of the effects of key operating variables in ETBE reactive distillation columns. Several useful results for ETBE synthesis via reactive distillation were obtained from the homotopic study of simulation columns: (1) Ethanol is predominantly recovered with the bottoms product in an ETBE reactive distillation column, whereas methanol is recovered with the distillate product from a MTBE reactive distillation column. (2) ETBE production is more attractive with low isobutylene feed concentrations, whereas MTBE production is best with around 60% isobutylene in the feed. (3) The numbers of rectification, reaction, and stripping section stages generally need to be optimized for maximum isobutylene conversion. (4) A high reflux rate is necessary for maximum isobutylene conversion, although satisfactory operation can be achieved with a low reflux rate and more stages. (5) The reboiler duty needs to be tightly controlled around a narrow optimum for satisfactory column operation. The results of the study were used to develop an integrated design method for reactive distillation columns for ether synthesis. The design method relies on accurate simulation results and an iterative process to locate an acceptable design meeting all the process requirements. Nomenclature ai ) activity of component i HL ) molar liquid enthalpy HV ) molar vapor enthalpy krate ) reaction rate constant KA ) ethanol adsorption equilibrium constant KDIB ) DIB reaction equilibrium constant KETBE ) ETBE reaction equilibrium constant

L ) molar liquid flow mcat ) mass of catalyst P ) pressure Pvap ) vapor pressure Qc ) condenser duty Qr ) reboiler duty r1,i ) reaction rate of component i in the first reaction (ETBE) r2,i ) reaction rate of component i in the second reaction (DIB) TL ) liquid temperature TV ) vapor temperature V ) molar vapor flow xi ) molar liquid concentration of component i yi ) molar vapour concentration of component i γi ) activity coefficient of component i Abbreviations DIB ) diisobutylene ETBE ) ethyl tert-butyl ether EtOH ) ethanol iBut ) isobutylene MON ) motor octane number MTBE ) methyl tert-butyl ether nBut ) normal butylenes (1-butylene and 2-butylene) RON ) research octane number RVP ) Reid vapor pressure

Literature Cited Abufares, A. A.; Douglas, P. L. Mathematical Modeling and Simulation of an MTBE Catalytic Distillation Process Using SpeedUp and Aspen Plus. Chem. Eng. Res. Des. 1995, 73, 3. ARCO Chemical Technology. Ethers, Petrochemical Processes ‘95. Hydrocarbon Process. 1995, 74 (3), 110. Aspen Technology Inc. The SpeedUp User’s Manual; Aspen Technology Inc.: Cambridge, MA, 1993. Bravo, J. L.; Pyhalahti, A.; Ja¨rvelin, H. Investigations in a Catalytic Distillation Pilot Plant: Vapor/Liquid Equilibrium, Kinetics, and Mass-Transfer Issues. Ind. Eng. Chem. Res. 1993, 32, 2220. Brockwell, H. L.; Sarathy, P. R.; Trotta, R. Synthesize Ethers. Hydrocarbon Process. 1991, 70 (9), 133. Columbo, F.; Corl, L.; Dalloro, L.; Delogu, P. Equilibrium Constant for the Methyl tert-Butyl Ether Liquid Phase Synthesis by Use of UNIFAC. Ind. Eng. Chem. Fundam. 1983, 22, 219. Dean, J. A. Lange’s Handbook of Chemistry, 14th ed.; McGrawHill: New York, 1992. Flato, J.; Hoffman, U. Development and Start-up of a Fixed Bed reaction Column for Manufacturing Antiknock Enhancer MTBE. Chem. Eng. Technol. 1992, 15, 193. Furzer, I. Australian Gasoline and MTBE. Chem. Eng. Aust. 1994, 19 (4), 9. Gmehling, J.; Onken, V. Vapor-liquid Equilibrium Data Collection; DECHEMA: Frankfurt, Germany, 1977. Gmehling, J.; Menke, J.; Krafczyk, J.; Fischer, K. Azeotropic Data, Part 1; VCH: Weinheim, Germany, 1994. Jensen, K. L.; Datta, R. Ethers from Ethanol. 1. Equilibrium Thermodynamic Analysis of the Liquid Phase Ethyl tert-Butyl Ether Reaction. Ind. Eng. Chem. Res. 1995, 34, 392. Jensen, K. L.; Datta, R. Ethers from Ethanol. 7. Transition State Theory Analysis of the Kinetics of Liquid Phase Ethyl tert-Butyl Ether Synthesis Reaction. Submitted to Ind. Eng. Chem. Res. 1996. Kister, H. Z. Distillation Design; McGraw-Hill: New York, 1992. Kitchaiya, P.; Datta, R. Ethers from Ethanol. 6. Kinetics of Simultaneous tert-Amyl Ethyl Ether Synthesis and Isoamylene Isomerisation. Submitted to Ind. Eng. Chem. Res. 1996. Kra¨henbu¨hl, M. A.; Gmehling, J. Vapor Pressures of Methyl tertButyl Ether, Ethyl tert-Butyl Ether, Isopropyl tert-Butyl Ether, tert-Amyl Methyl Ether and tert-Amyl Ethyl Ether. J. Chem. Eng. Data 1994, 39, 759. Lide, D. R., Ed. CRC Handbook of Chemistry and Physics, 75th ed.; CRC: Boca Raton, FL, 1994. Miracca, I.; Tagliabue, L.; Trotta, R. Multitubular Reactors for Etherifications. Chem. Eng. Sci. 1996, 51 (10), 2349.

Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1869 Piel, W. J.; Thomas, R. X. Oxygenates for Reformulated Gasoline. Hydrocarbon Process. 1990, 69 (7), 68. Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Properties of Gases and Liquids, 4th ed.; McGraw-Hill: New York, 1987. Sarathy, P. R.; Suffridge, G. S. Etherify Field Butanes (Part 2). Hydrocarbon Process. 1993, 72 (2), 43. Simulation Sciences Inc. Pro/II Keyword Input Manual; Simulation Sciences Inc.: Brea, CA, 1994. Simulation Sciences Inc. Pro/II Casebook. Methyl Tertiary Butyl Ether (MTBE) Plant; Simulation Sciences Inc.: Brea, CA, 1995. Smith, L. A. Catalytic Distillation Process and Catalyst. Eur. Pat. Appl. EP8860, 1980. Sneesby, M. G.; Tade´, M. O.; Datta, R. tert-Butyl EtherssA Comparison of Properties, Synthesis Techniques and Operating Conditions for High Conversions. Dev. Chem. Eng. Miner. Process. 1995, 3, 89. Sundmacher, K.; Hoffmann, U. Oscillatory Vapor-Liquid Transport Phenomena in a Packed Bed Reactive Distillation Column for Fuel Ether Production. Chem. Eng. J. 1995, 57, 219. Ung, S.; Doherty, M. F. Synthesis of Reactive Distillation Systems with Multiple Equilibrium Chemical Reactions. Ind. Eng. Chem. Res. 1995, 34, 2555.

Unzelman, G. H. Ethers in Gasolines1. Ethers Have Good Gasoline Blending Attributes. Oil Gas J. 1989, Apr 10, 33. Zhang, T.; Datta, R. Integral Analysis of Methyl tert-Butyl Ether Synthesis Kinetics. Ind. Eng. Chem. Res. 1995, 34, 730. Zhang, T.; Kitchaiya, P.; Jensen, K. L.; Phillips, C.; Datta, R. Ethers from Renewable/Recoverable Resources. Final Technical Report NREL/#XAT-3-13413-01; University of Iowa: Iowa City, IA, 1995.

Received for review May 20, 1996 Revised manuscript received January 15, 1997 Accepted January 16, 1997X IE960283X

X Abstract published in Advance ACS Abstracts, February 15, 1997.

ETBE Synthesis via Reactive Distillation. 1. Steady-State Simulation

Recommend Documents