Simulation of Hydrogen Production from Dehydrogenation of Ethanol in a Palladium Membrane Reactor

Transcription

1 J. Chin. Inst. Chem. ngrs., Vol. 33, o. 3, , 22 Simulation of ydrogen roduction from Dehydrogenation of thanol in a alladium Membrane Reactor Kuang-Yi Chang [1], Wen-siung Lin [2] and sin-fu Chang [3] Department of Chemical ngineering, Feng Chia University Taichung, Taiwan 47, R.O.C. bstract In this study, we have employed a Runge -Kutta numerical method to simulate the hydrogen production from the dehydrogenation of ethanol in a palladium membrane reactor. The reacting system consists of a double-tube reactor with a palladium membrane coated on the inner porous stainless steel tube. The shell side is loaded an appropriate amount of active catalyst over which the dehydrogenation of ethanol occurs. n empirical rate law from Franckaerts and Froment (1964) is used to describe the ethanol dehydrogenation behavior. Onedimensional plug flow, negligible axial dispersion of heat and mass transfer, and negligible radial gradients of temperature and concentration are assumed before conducting the numerical simulation. The simulation results show that the membrane reactor, which combines reaction and separation in a single unit, not only increases the reaction conversion, but also improves the efficiency of the reactor. When the ethanol feed is mole/s and with a purge gas flow rate of 1-3 mole/s in the permeation side, the conversion will achieve 1% which exceed the equilibrium value of 81.7%. With proper operating conditions, 1% recovery of hydrogen can be achieved either by using a high flow rate of purge gas in the separation side or by reducing the pressure of separation side to nearly vacuum. Key Words : alladium membrane reactor, thanol dehydrogenation, Runge-Kutta simulation ITRODUCTIO ydrogen becomes a potential substitute fuel because petroleum faces its dramatic shortage in the near future. Scientists all over the world strive to develop new processes to produce concentrated hydrogen more efficiently and economically. Conventionally, hydrogen is made in large scale by the steam reforming of hydrocarbons such as methane or naththa oil. It also exists as a by-product from several chemical processes such as the production of caustic soda from electrolysis of brine, coal gasification, etc. Recently, global environmental requirements have motivated the researches on optimal treatment of biomass and hydrogen production from it has become a main subject (Shiga et al., 1998). Membrane reactors are processing chemical reaction and separation in one unit. The membrane, being selective to one or more of the product or reacting species, can be used to improve the yields of thermodynamically limited reactions (Gobina and ughes, 1994). When conducting dehydrogenation reactions, which are normally endothermic, palladium membrane reactors appear promising since the selective removal of hydrogen would inevitably tend to shift the equilibrium toward increased product yields. In the past twenty years palladium-based membranes were tested in the numerous reactions with removal or supply of hydrogen to packed-bed reactors (Dittmeyer et al., 21), these reactions include dehydrogenation of cyclohexane to benzene (Itoh, 1987, 1995), dehydrogenation of propane to propylene (Weyten et al., 2), dehydrogenation of ethylbenzene to styrene (Quicker et al., 2), etc. owever, only a few of researchers have shed light on the ethanol dehydrogenation using palladium membrane reactor (mandusson et al., 21; Cao et al., 1997). In this communication we are going to simulate hydrogen production from the dehydrogenation of ethanol in a palladium membrane reactor and further experimental work is being conducted in our lab to verify the simulation results. thanol can be easily obtained elsewhere through the fermentation of crops and after dehydrogenation its main product acetaldehyde is an invaluable important intermediate chemical serving as the raw material for the production of acetic acid, acetic anhydride and many other products. owever, it is difficult to get [1] [2] [3], To whom all correspondence should be addressed

2 226 J. Chin. Inst. Chem. ngrs., Vol. 33, o. 3, 22 Fig. 1. Flow model in a palladium membrane reaction. dehyde in high concentration since the conversion of the direct dehydrogenation process is relatively low due to the thermodynamic equilibrium limitation. Using the membrane reactors can significantly improve the conversion of ethanol and consequently obtain concentrated acetaldehyde and hydrogen (Deng et al., 1995; Cao et al., 1997). MODLIG The dehydrogenation of ethanol C3 C 2O C3CO + 2 occurs in the shell side of two concentric tubes with radii, R i and R, respectively, in which an active dehydrogenation catalyst is packed. The palladium membrane is coated on the inner porous stainless steel pipe and has an axial length of 5cm. thanol is fed into the reactor with nitrogen as a carrying gas ( u / u =4.5) (Chou, 1996). The product hydrogen permeating through the membrane into the permeation side can be collected either by using a purge gas or by evacuation. Figure 1 shows the membrane reactor in which the flow pattern of each component gas is indicated. In this study the following assumptions have been made: (1) One-dimensional plug flow. (2) egligible axial dispersion of heat and mass transfer. (3) egligible radial gradients of temperature and concentration. set of governing equations describing the reactor can be formulated by taking a material balance of each component in a differential mass of catalyst. Therefore, for ethanol, the mole balance equation is u x u + r W =, (1) x+ x 2 2 i b r b where W = π( R R ) xρ = xρ, = π( R r Q = 2 2 Ri ). Dividing each term of q. (1) by x and then taking the limit of x to zero, we have the differential form du / dx = r ρ (2) r b The rate of dehydrogenation of ethanol is given by (Franckaerts and Froment, 1964) r = kk ( /(1 + K + K / K where lnk = 1615/RT+16.14, lnk = 371/RT 4.2, lnk = 596/RT 4.73, lnk = G r /RT. ) 2 ), (3) The change in Gibbs free energy, G r, can be calculated as follows (Yaws and Chiang, 1988) G where G r = G products G reactants = G + G G = T T 2, G =, G = T T 2. ow, the mole balance on hydrogen u x u 2 π Ri x Q + r W =, (4) x+ x where Q is the permeation rate of hydrogen gas through the palladium membrane and according to Sieverts law α ( ), (5) where α, the permeance of the palladium membrane, is mol/cm 2 s (Lin and Rei, 2).

3 Kuang-Yi Chang, Wen-siung Lin and sin-fu Chang : Simulation of ydrogen roduction from Dehydrogenation of 227 thanol in a alladium Membrane Reactor Since r = r, we have du dx = r r ρb 2πRiα ( ). (6) The partial pressures of hydrogen in the reaction side and the permeation side, / and /, can be represented respectively as / = Tr = Tr u u Σ u ( u i + u + u + u / Ts i Ts = υ Συ = υ /( υ + υ ) ) where Tr = r / and Ts = s / then we have du dx = r ( ρ r Tr b 2π R α u u Other conservation correlations are u i i Ts υ ). υ i (7) = u u (8) Fig. 2. The equilibrium conversion of ethanol dehydrogenation in a conventional reactor. υ = u ) u (9) ( u u = u, υ = υ, υ = υ = (constant). quations (2), (7) (9) were numerically solved by the Runge-Kutta method with the following initial conditions at x =, = u, u =, u =, u = u, u υ = υ, υ = and using the following data : L = 5 cm, ρ b =.5 g/cm 3, = 1 atm, R = 1.5 cm, R i =.625 cm, u = 4.5 u mol/s. (a) RSULTS D DISCUSSIO The dehydrogenation reaction follows a stoichiometric relation: C 3 C 2 O C 3 CO + 2 with the equilibrium constant K p =exp( G r /RT). Figure 2 shows the variations of equilibrium conversion with respect to pressure and temperature, in accordance with thermodynamic restrictions. The conversion increases with increasing temperature and decreasing pressure, which is obliged by the Le Chatelier s principle for an endothermic decomposition reaction. vidently, the conversion can be enhanced with increasing reaction temperatures and reducing pressure, but doing so is not always (b) Fig. 3. The flow rate of each component in a palladium membrane reactor as a function of (a)reactor length; (b)contact time. economically feasible. Figure 3 shows the flow rate of each compo-

4 228 J. Chin. Inst. Chem. ngrs., Vol. 33, o. 3, 22 (a) (b) (c) Fig. 4. ffect of nitrogen flow rates in the permeation side on the ethanol conversion ; reaction pressure : (a) 1 atm; (b) 5 atm; (c) 1 atm. Fig. 5. The variation of conversion with respect to nitrogen flow rates at different s. nent in the reaction side of the palladium membrane reactor at various reactor lengths and space time. Since the catalyst is very active the ethanol consumes very rapidly with accompanying of acetaldehyde as well as hydrogen production and the reaction is almost completed at reactor length of 1 cm. In a conventional catalytic reactor hydrogen is produced at the same rate as acetaldehyde and this also holds for a membrane reactor. t the entrance region of reactor the hydrogen production rate is faster than its permeation rate through the membrane, which results an accumulation of hydrogen concentration and the hydrogen flow rate increases with reactor length. fter a short distance from the entrance the reaction rate drops rapidly since most of the ethanol molecules are consumed, which causes hydrogen production rate to be much lower than the permeation rate, resulting a reduction of hydrogen flow rate in the reaction side and creating a maximum of it as shown in Fig. 3. Figure 4 shows that at reaction temperature of 3 C and different reaction pressures the conversion increases with increasing flow rates of purge gas. t a fixed flow rate of ethanol it can be seen that an increasing flow rate of nitrogen in the permeation side results in higher conversion, a remarkable promotion of the reaction especially when the membrane reactor is operated at higher pressures. li et al. (1994) experimentally obtained similar results through the conducting the dehydrogenation of methylcyclohexane to toluene in a d-g membrane reactor under the pressure range of 5 2 bars. s the flow rate of purge gas increases, the hydrogen partial pressure in the permeation side decreases due to a dilution effect and the pressure difference across the membrane becomes larger, which therefore cause a higher hydrogen flux according to the Sieverts law. When the flow rate of purge gas is fixed, increasing

5 Kuang-Yi Chang, Wen-siung Lin and sin-fu Chang : Simulation of ydrogen roduction from Dehydrogenation of 229 thanol in a alladium Membrane Reactor Fig. 6. The temperature effect on the hydrogen flow rate in the permeation side. feed rate, i.e., decreasing the residence time of ethanol in the reactor, results in a reduction in conversion. Itoh (1987) also found that in a palladium membrane reactor the conversion of cyclohexane to benzene can be easily increased from an equilibrium value of 18.7% to nearly 1% at a sufficiently large residence time and by using enough sweep gas. Under atmospheric pressure, when there is no purge gas in the permeation side, i.e. υ =, the conversions will approach a limiting value, 81.7%, which is the equilibrium conversion, no matter what the ethanol flow rate is. Because hydrogen pressure in the permeation side will build up as the nitrogen flow is ceased and eventually the hydrogen partial pressures of both sides will be equal, which results in no flux of hydrogen across the membrane. The membrane reactor behaves itself as a conventional reactor. The conversion reaches 99.8% when u = mole/s and υ = mole/s, which is much higher than the equilibrium conversion, 81.7%, the maximum conversion achieved by using a conventional catalytic reactor. With proper operating conditions, the membrane reactor is expected to give 1% conversion and which means only acetaldehyde leaves the outlet of the reaction side, no additional equipment is needed to separate ethanol and acetaldehyde and the operating cost is apparently reduced. When the permeation side is subject to evacuation, the hydrogen flux becomes larger due to a higher difference of partial pressures across the membrane and the conversion also increases with the increasing extent of evacuation, as shown in Fig. 5. The hydrogen flow rate in the permeation side increases with increasing temperature as indicated in Fig. 6. It is found that υ is weakly dependent on temperature when T K. t very low feed rate, there still exists a temperature effect as can be seen from the inset in Fig. 6. Recovery of hydrogen, defined as number of moles of hydrogen extracted from the separation side per mole of hydrogen produced in the reaction side, varying with ethanol flow rates at specified flow rates of purge gas is shown in Fig. 7. It decreases with increasing ethanol flow rate because the residence time decreases. It is interesting to note that the recovery of hydrogen also increases with increasing flow rates of purge gas, owing to a higher hydrogen pressure gradient across the membrane. Recovery of -6 hydrogen can easily attain 99% when u = 1 mole/s and υ = mole/s, indicating that a direct hydrogen production in the course of reaction would be possible. Figure 8 shows that hydrogen flow rate in the permeation side at a specific ethanol feed rate remarkably increases with increasing flow rate of purge gas and levels off with a further increase. This tells us that with an appropriate amount of purge gas a production rate of hydrogen can be easily achieved. The reaction pressure effect on both

6 23 J. Chin. Inst. Chem. ngrs., Vol. 33, o. 3, 22 υ and recovery of hydrogen is shown in Fig. 9. The hydrogen production rate increases with increasing ethanol feed, and is accelerated at a high reaction pressure. On the contrary, the recovery of hydrogen attenuates with increasing reactant feed, but also increases with increasing reaction pressure. Instead of using a purge gas to speed up the permeation rate of hydrogen as described above, there is an alternative method for removing the permeated hydrogen simply by maintaining the pressure of separation side at a reduced level or vacuum. Figure 1 shows the effect of reaction side pressure on the ethanol conversion when the permeation side is subjected to evacuation ( s =.1 atm). reversible reaction will behave like an irreversible reaction when there is only a trivial amount of product to boost the reverse reaction. s the ethanol flow rate exceeds mole/s, Fig. 1 shows that the conversion is higher for a higher reaction side pressure, Fig. 7. Dependence of recovery of hydrogen on flow rate of purge gas. Fig. 9. The reaction pressure effect on υ and recovery of hydrogen. Fig. 8. ydrogen flow rate in the permeation side at different nitrogen flow rates and ethanol feed rates. Fig.1.ffect of reaction side pressure on the ethanol conversion when the permeation side is subjected to evacuation. contradicting the prediction of Le Chatelier s principle. This indicates a higher reactant pressure resulting a higher reaction rate. t a fixed flow rate of reactant the recovery of hydrogen increases with decreasing s is depicted in Fig. 11. It reaches almost 1% when u = 1-5 mole/s and s =.1 atm, indicating that a highly concentrated hydrogen can be obtained from the separation side with further purification. In conclusion we have simulated the dehydrogenation of ethanol in a palladium reactor with operating conditions in compliance with those published in the literature. The simulation data will be very helpful to providing an aid in experimental work, which is being conducted in our lab, on avoiding unnecessary tests and explicating invaluable results.

7 Kuang-Yi Chang, Wen-siung Lin and sin-fu Chang : Simulation of ydrogen roduction from Dehydrogenation of 231 thanol in a alladium Membrane Reactor Fig.11.Recovery of hydrogen at different ethanol flow rates and permeation side pressures. CKOWLDGMT The authors are indebted to Dr. Y.-M. Lin of ITRI for his generous endowment of palladium membrane tubes that make the preliminary experimental work successful in this lab. The authors also like to thank the ational Science Council of the Republic of China for the financial support. (Grant o. SC C-35-9). OMCLTUR r cross-sectional area of annular region, cm 2 G r the change of Gibbs free energy, KJ/mol G i Gibbs free energy of formation of component i, KJ/mol -1 k apparent rate constant, mol g-cat -1 s K adsorption equilibrium constant of acetaldehyde, atm -1 K adsorption equilibrium constant of ethanol, atm -1 K equilibrium constant, atm L dimensionless length of reactor L entire length of reactor, cm partial pressure of hydrogen on reaction side, atm partial pressure of hydrogen on permeation side, atm i partial pressure of component i on reaction side, atm reference pressure, = 1 atm r total pressure on reaction side, atm s total pressure on permeation side, atm Tr dimensionless total pressure on reaction side (= r / o ) Ts dimensionless total pressure on permeation side (= s / o ) permeation rate of hydrogen through Q -2 membrane, mol s -1 cm r rate of ethanol dehydrogenation, mol gcat -1-1 s -1 r rate of hydrogen formation, mol g cat -1 s R i radius of inner tube, cm R radius of outer tube, cm T absolute temperature, K u flow rate of ethanol at reaction-side inlet, mol s -1 u flow rate of nitrogen at reaction-side inlet, mol s -1 u i υ υ i W x Greek symbols α flow rate of gas i in reaction-side stream, mol s -1 flow rate of nitrogen at permeation-side inlet, mol s -1 flow rate of gas i in permeation-side -1 stream, mol s total weight of catalyst bed, g axial coordinate, cm permeation rate constant of hydrogen, mol s -1 cm -2 ρ density of catalyst bed, g cm -3 b Subscripts acetaldehyde ethanol hydrogen nitrogen RFRCS li, J. K.,. J ewson. and D. W. Rippin, xceeding quilibrium Conversion with a Catalytic Membrane Reactor for the Dehydrogenation of Methylcyclohexane, Chem. ng. Sci., 49, 2129 (1994). mandusson,., L.-G. kedahl and. Dannetun, lcohol Dehydrogenation over d Versus dg Membranes, ppl. Catal., 217, 157 (21). Cao, Y., B. Liu and J. Deng, Catalytic Dehydrogenation of thanol in d-m/γ-l 2 O 3 Composite Membrane Reactors, ppl. Catal., 154, 129 (1997). Chou, Y.., The ffect of alladium on Copper Dispersion and thanol Dehydrogenation ctivity of Cu/l 2 O 3 Catalyst repared by the lectroless lating rocedure, M.S. Thesis, Department of Chemical ngineering, Feng Chia University, Taichung, Taiwan (1996). Deng, J., Z. Cao and B. Zhou, Catalytic Dehydrogenation of thanol in a Metal-Modified lumina Membrane Reactor, ppl. Catal., 132, 9 (1995). Dittmeyer, R., V. ollein and K. Daub, Membrane Reactors for ydrogenation and Dehydrogenation rocesses Based

8 232 J. Chin. Inst. Chem. ngrs., Vol. 33, o. 3, 22 on Supported alladium, J. Mole. Catal. : Chemical, 173, 135(21). Franckaerts, J. and G. F. Froment, Kinetic Study of the Dehydrogenation of thanol, Chem. ng. Sci., 19, 87 (1964). Gobina,. and R. ughes, thane Dehydrogenation Using a igh-temperature Catalytic Membrane Reactor, J. Membr. Sci., 9, 11 (1994). Itoh,., Membrane Reactor Using alladium, ICh J., 33, 1576 (1987). Itoh,., Limiting Conversions of Dehydrogenation in alladium Membrane Reactors, Catal. Today, 25, 351(1995). Lin, Y.-M. and M.-. Rei, rocess Development for Generating igh urity ydrogen by Using Supported alladium Membrane Reactor as Steam Reformer, Int. J. ydrogen nergy, 25, 211(2). Quicker,., V. ollein and R. Dittmeyer, Catalytic Dehydrogenation of ydrocarbons in alladium Composite Membrane Reactors, Catal. Today, 56, 21(2). Shiga,., K. Shinda, K. agiwara,. Tsutsumi, M. Sakurai, K. Yoshida and. Balgen, Large-Scale ydrogen roduction from Bomass, Int. J. ydrogen rod., 21, 631 (1998). Weyten,., J. Luyten, K. Keizer, L. Willems and R. Leysen, Membrane erformance: the Key Issues for Dehydrogenation Reactions in a Catalytic Membrane Reactor, Catal. Today, 56, 3 (2). Yaws, C. L and -Y. Chiang, Find Favorable Reactions Faster, ydrocarb rocess, 81 ov. (1988). (Manuscript Received September 29, 21) Runge-Kutta (shell side) Franckaerts Froment (1964) (plug flow) (dispersion) mol/s (purge gas) 1-3 mol/s 1% 81.7% (separation (purge side) gas) 1%

9 2 J. Chin. Inst. Chem. ngrs., Vol. 33, o.2, 22 acetal-