Ind. Eng. Chem. Res. 2006, 45, 1009-1013
1009
Nonpremixed Catalytic Combustion of Methane in a Fluidized Bed Reactor Mario Iamarino,*,† Paola Ammendola,‡ Riccardo Chirone,§ Raffaele Pirone,§ Giovanna Ruoppolo,§ and Gennaro Russo§ Dipartimento di Ingegneria e Fisica dell’Ambiente, UniVersita` degli Studi della Basilicata, Via dell’Ateneo Lucano 10, 85100 Potenza, Italy, Dipartimento di Ingegneria Chimica, UniVersita` Federico II, Piazzale Tecchio, 80125 Napoli, Italy, Istituto Ricerche sulla CombustionesCNR, Piazzale Tecchio, 80125 Napoli, Italy
The catalytic combustion of methane has been investigated in a 0.10 m bubbling fluidized bed reactor with nonpremixed feedings of reactants. Copper dispersed on porous γ-Al2O3 spheres (1 mm diameter) characterized by high mechanical strength has been used as catalyst. The effect of design variables such as methane inlet concentration, bed temperature, and superficial gas velocity on methane conversion has been quantified in the ranges 4-10 vol %, 650-750 °C, and 0.40-1.30 m s-1, respectively. The propensity to attrition of the catalyst has been separately investigated under the experimental conditions tested. Results have been interpreted in the light of a simple reactor model which assumes a plug-flow pattern for the gas through the bed and mth-order catalytic kinetics with respect to fuel concentration. 1. Introduction Fluidized bed catalytic converters represent a viable alternative to packed beds and monoliths for catalytic combustion applications, mainly due to improved heat transfer properties, enabling, on one hand, an efficient heat recovery by means of external or submerged heat transfer surfaces and preventing, on the other, undesired catalyst overheating1 which could result in its thermal deactivation. In addition, catalytic fluidized bed fed with gaseous, liquid, or solid fuel has revealed very attractive features when compared with traditional oxidation systems, mainly due to the lower content of pollutants in the exhaust gas and the higher volumetric density of heat produced.2 Despite these favorable considerations, in the past fluidized bed catalytic converters have not received enough attention and open issues, such as the loss of expensive catalyst due to attrition4 and the criticality of fuel/air mixing phenomena in fluidized beds operated in the bubbling regime,3 still pose technical barriers to the full exploitation of the fluidized bed catalytic conversion technology. Actually, the possibility of operating the reactor in the turbulent fluidization regime,5 advocated as a way to overcome the onset of bubble-to-emulsion phase mass transfer resistances, results in larger losses of catalyst as attritted material due to the higher gas superficial velocities. It follows that additional research efforts are required to find an optimal tradeoff between these conflicting requirements: setup of a catalyst with low attrition propensity and careful optimization of hydrodynamics and gas mixing regimes of fluidized catalytic converters. According to previous studies,6-8 a catalytic system based on copper supported on porous γ-alumina spheres has been proven to display good activity, low cost, and excellent thermal and mechanical stability for use in fluidized bed catalytic combustion of light hydrocarbons. The papers refer to a study on catalytic combustion of methane and propane in a premixed bubbling fluidized bed under fuel-lean conditions and highlight how the use of coarse catalyst pellets could favorably affect * To whom correspondence should be addressed. E-mail: iamarino@ unibas.it. Tel.: +39 0971205208. Fax: +39 0971205160. † Universita` degli Studi della Basilicata. ‡ Universita` Federico II. § Istituto Ricerche sulla Combustione.
the effectiveness of bubble-to-emulsion phase mass transfer, with complete and stable hydrocarbon conversion obtained at temperatures below 700 °C; propensity to attrition and thermal deactivation of the catalyst turned out to be fairly low. Experimental results have been interpreted on the basis of competing phenomena between intraparticle and interphase diffusional processes and intrinsic kinetics of heterogeneous reaction. With reference to the premixed option, the boundary layer diffusion around the catalyst particle turns out to be faster than the other processes, while the competition between intrinsic reaction kinetics, intraparticle diffusion, and bubble-to-emulsion phase mass transfer controls the converter performance under the range of temperatures investigated. The present work moves further to address the performance of catalytic converters characterized by nonpremixed feedings of methane and air, relevant to the combustion of mixtures within the flammability limits. This extension is of increasing interest with reference to practical applications of gas fluidized bed boilers when considering the high potentiality required and the safety problems related to a flammable feeding. Effects of methane inlet concentration, gas superficial velocity, and temperature on reactor performance have been investigated in terms of fuel conversion, unburned fuel emissions, and particulate elutriation rate. In addition, the effect of methane inlet concentration on conversion has been interpreted in the light of a simple reactor model that assumes a plug-flow pattern for the gas through the bed and mth-order catalytic kinetics with respect to fuel concentration. 2. Experimental Section The experimental apparatus used in this work is reported in Figure 1. The catalytic converter consisted of a 0.10 m stainless steel fluidized bed reactor, equipped with a porous plate as air distributor. Methane was injected in the bed 2 cm above the porous plate through a distributor consisting of a cross-shaped manifold, with four branches 4 cm long. Four holes with 0.8 mm diameter were drilled along each branch, equally spaced and perpendicularly oriented with respect to the reactor axis. On the basis of literature studies,9 the discharge mode of methane at the distributor level is expected, at the experimental conditions adopted, to give rise to stationary horizontal jets penetrating into the bed rather than to trains of single ascending
10.1021/ie051015e CCC: $33.50 © 2006 American Chemical Society Published on Web 12/27/2005
1010
Ind. Eng. Chem. Res., Vol. 45, No. 3, 2006
Figure 1. Experimental setup.
bubbles, but the setup did not allow a direct visual observation of the bed to verify the discharge regime. Below the porous plate, a packed bed (0.60 m height) consisting of Rasching rings of different sizes was used to level off the inlet air velocity profile. An electrical oven surrounding the packed bed was employed during startup to initially heat the reactor to about 400 °C. At this temperature, bed activity was high enough to start the reaction and ensure further temperature increase. During steady-state operation, the oven was switched off and the temperature was stabilized by removing the excess of heat generated by means of a cooling system consisting of a copper coil wound up around the external reactor surface for an height of 0.15 m. Liquid water was used as exchange medium, subtracting heat by complete vaporization; vapor was then condensed and recirculated to the reactor. Temperature profiles along the reactor axis were measured by means of five equally spaced thermocouples (T1-T5 in Figure 1) vertically inserted into the bed and located at fixed positions (0.06, 0.145, 0.23, 0.315 and 0.4 m, respectively, above to the air distributor). Continuous monitoring of gas composition at the exhaust line was accomplished by on-line nD-IR analyzers (for CH4, CO, CO2, O2, and NOx) after water removal. A data acquisition unit was used to log on signals from the experimental facility, and concentration measurements were properly corrected to take into account the removal of water. The bed consisted of 2100 g of catalyst with corresponding unexpanded bed height of 0.30 m. The catalyst was obtained by dispersing copper on porous γ-Al2O3 spheres supplied by Sasol, characterized by high mechanical strength and lying at the borderline between groups B and D of the Geldart classification of powders.10 Catalyst was thermally pretreated at 800 °C to yield copper mainly as superficial spinel phase CuAl2O4, stable upon repeated thermal treatments. Preparation techniques and physicochemical properties of catalyst are detailed in a previous work.6 Methane conversion experiments were carried out at methane inlet concentrations mostly ranging from 4 to 9 vol %, while a few tests were performed up to 10 vol %, hence above the
Figure 2. Axial temperature profiles measured at Cin ) 6 vol % and u0 ) 0.40 (b), 0.60 (9), 0.80 (2), and 1.0 (1) m s-1. Dotted line indicates unexpanded bed height.
stoichiometric value (9.5 vol %) for methane in air. The temperature range investigated was between 650 and 750 °C, where the upper limit for temperature is consistent with the catalyst thermal deactivation threshold (800 °C). Catalyst mechanical stability was assessed at different temperatures, methane inlet concentrations, and superficial velocities by collection of solid particles in the exhaust gas by means of an isokinetic withdrawing system, since high pressure drops would be reached by filtering of the total flow. This system consists of a 1 mm stainless steel probe, coaxially inserted with respect to the fluidized bed reactor and located 0.2 m below its upper end, a thermocouple (T6 in Figure 1) for temperature measurement of the exhaust gas at the collecting point, a ceramic filter able to collect fines larger than 300 nm, and a pump able to extract a desired gas flow. 3. Results and Discussion Figure 2 reports typical temperature profiles measured at fixed axial positions in experiments at 700 °C and 6 vol % methane inlet concentration for different values of the fluidization velocity
Ind. Eng. Chem. Res., Vol. 45, No. 3, 2006 1011
Figure 3. Cout/Cin as a function of u0/umf at temperatures of 650 (b), 700 (9), and 750 °C (2). For each temperature, data correspond to inlet methane concentrations in the range 4-9 vol %.
u0. For the lowest value of u0 (0.40 m s-1), the highest thermocouple (at 0.40 m above air distributor) hangs outside the bed due to the limited bed expansion and the corresponding temperature is remarkably lower. Increasing u0, all thermocouples became immersed in the bed and temperature differences were smaller, according to the improved solid mixing. In more detail, while temperature differences up to 10 °C have been observed at u0 ) 0.40 and 0.60 m s-1, at 1.0 m s-1 they do not exceed 1-2 °C. In any case the higher temperatures were always measured in the lower section of the bed, where most of the heat release takes place. Due to small but possible temperature differences, data reported hereafter have been always conventionally referred to the temperature measured by the lowest thermocouple in the bed, indicated as T1 in Figure 1 and located 0.06 m above the air distributor. It can be also observed that, since thermocouples are directly in contact with the solid phase, the values measured are mostly indicative of the temperature of the catalyst external surface. During combustion tests, the reactor was characterized by very high methane conversion degrees (mostly above 0.99) and steady-state operation in the autothermal regime. For every test at inlet fuel concentration below the stoichiometric value (9.5 vol %), analysis of the exhaust gas only revealed products of complete methane oxidation without any detection of CO or NOx. This evidence indicates that oxidation takes place entirely on the catalyst surface and that homogeneous reaction, which is a well-known source for the formation of CO and NOx in hot fluidized bed of inert11 and catalytic12 particles (where the reactive mixture can ignite in bubble phase or even in the freeboard region), is not relevant under the experimental conditions investigated. Moreover, O2, CO2, and CH4 amounts measured at the outlet were always consistent with the corresponding mass balances. A few tests performed at a fuel concentration of 10 vol %, hence above the stoichiometric value, also revealed the formation of very small amounts of CO (20 and 50 ppm, respectively, at 700 and 750 °C, with u0 ) 0.60 m s-1), while fuel conversion accordingly decreased due to oxygen shortage. Figure 3 reports methane concentration measured at the outlet normalized with respect to inlet concentration, Cout/Cin, as a function of the ratio between gas superficial velocity and the minimum fluidization velocity, u0/umf. According to previous findings,8 values of 0.21, 0.20, and 0.19 m s-1 have been considered for umf at temperatures of 650, 700, and 750 °C, respectively. Experimental data corresponding to the same bed
Figure 4. Outlet methane concentration [ppm] at high conversion degrees (Cin ) 9 vol %, T ) 700 (9) and 750 °C (2)).
Figure 5. Normalized outlet methane concentration at Cin ) 6 vol % and u0 ) 0.40 (b), 0.60 (9), and 0.80 (2) m s-1.
temperature have been reported with the same symbol whatever the methane inlet concentration (in the range 4-9 vol %). Analysis of the figure shows that there is a range of gas velocities, u0/umf < 2.5, where full conversion of methane (conversion degrees higher than 0.999) has been achieved regardless of reactor temperature. At u0/umf > 2.5, unburned methane increases and this effect is larger when the operative temperature is decreased. It is worthwhile to observe that, in any case, a very small amount of methane were detected in the exhaust gas. As shown in Figure 4, the methane outlet concentration remains below approximately 100 ppm for u0/umf < 3.0 at 700 °C and for u0/ umf < 3.4 at 750 °C in accordance with the highest methane inlet concentration (9 vol %). On this basis, a maximum thermal power of about 4.5 kW can be produced by the converter at 750 °C with respect to constraints limiting fuel emissions to 100 ppm. In Figure 5, experimental results are reported in terms of normalized outlet methane concentration with respect to reactor temperature for three different gas superficial velocities (0.40, 0.60, and 0.80 m s-1) and constant methane inlet concentration (6 vol %). At 0.40 m s-1, almost complete methane conversion is reached in the whole temperature range investigated and a negligible effect of reactor temperature is recognized. The increase of gas velocity decreases methane conversion (increasing Cout/Cin), according to a decrease of the residence time of the gas in the reactor. This effect can be partially compensated
1012
Ind. Eng. Chem. Res., Vol. 45, No. 3, 2006 Table 1. Fines Elutriation Rate at Different Superficial Velocities
Figure 6. Effect of methane inlet concentration on Cout/Cin at T ) 750 °C and u0 ) 0.60 (b), 0.70 (9), and 0.80 (2) m s-1. Continuous lines have been obtained by the best fit of eq 1.
by an increase in reactor temperature: with respect to 0.40 m s-1, comparable reactor performances can be reached at 0.60 m s-1 for an operative temperature of 750 °C. The favorable effect of temperature is less effective at 0.80 m s-1, where the methane outlet concentration curve does not level off to zero in the range of temperature of practical interest for the catalyst used. Figure 6 reports Cout/Cin as a function of the inlet fuel concentration in experiments performed at 750 °C and u0 ) 0.60, 0.70, and 0.80 m s-1. The analysis has been restricted to gas velocities larger than 0.6 m s-1 according to the evidence that, resulting from Figure 3, the effect of Cin on methane conversion is not appreciable in the low velocity range. Analysis of data shows that, for all values of u0, unconverted methane increases when the inlet fuel concentration is increased. The effect of Cin has been analyzed in the light of a simple mathematical model of the reactor. A primary approximation of the catalytic converter is represented by the assumption of a plug-flow pattern for the gas through the bed (as suggested for fluidized beds of large particles13) and a reaction rate of mth order with respect to fuel concentration. Under these hypotheses, a mass balance on methane yields
Cout ) [1 + Da(m - 1)]1/(1-m) Cin
(1)
The Damko¨hler number, Da, is defined by
Da )
kVcat m-1 C Q in
(2)
where Vcat is the volume of catalyst, Q is the total gas flow rate, and k is an apparent kinetic constant. This model does not take into account the detailed mechanisms ruling the effective fuel conversion, and it lumps into both parameters k and m the effects of the intrinsic kinetics and of transport phenomena. Equation 1 was used to fit experimental data in Figure 6 with Da and m considered as unknown parameters. Good agreement with model prediction was found with m ) 0.94 at every u0, while best-fit Damko¨hler numbers showed an inverse proportionality with respect to the three gas superficial velocities, consistent with eq 2. The estimated value of the reaction order, 0.94, is intermediate between m ) 0.7, which corresponds to the intrinsic reaction order of methane oxidation on the catalyst
u0 [m s-1]
elutriation rate [g h-1]
0.4 0.6 0.8 1.0
not detectable 0.3 1.6 4.5
used,6 and m ) 1, corresponding to first-order phenomena with respect to methane concentration, as transport phenomena. This finding indicates that, in the experimental conditions investigated, the converter is not operating under purely kinetic control and that transport phenomena strongly affect reactor performance. To reinforce this conclusion, an apparent activation energy has been calculated on the basis of experimental data already reported in Figure 5 and obtained at u0 ) 0.60 and 0.80 m s-1 and different temperatures (data at u0 ) 0.40 have been excluded since the temperature effect is here extremely weak). By fitting of these data using eq 1 and m ) 0.94, apparent activation energies of 3900 and 4900 cal mol-1 were calculated respectively at u0 ) 0.60 and 0.80 m s-1. These values are considerably lower than the value of 29 800 cal mol-1 of the intrinsic catalytic kinetics7 and support the evidence of important mass transfer limitations on methane conversion. This conclusion is in good agreement with what was found in the premixed mode,7 but in this case the higher complexity of the mixing phenomena at the methane distributor level and the lack of a detailed reactor model do not allow any further conclusion on the specific nature of the controlling mechanisms. A first assessment of catalyst mechanical stability was carried out in ad hoc tests. The elutriation rates ware calculated by weighting the solids collected, and the resulting values showed a relatively strong dependency on the fluidization velocity but no relationship with respect to fuel inlet concentration and temperature. As reported in Table 1, an undetectable amount of fines was collected at u0 ) 0.40 m s-1, while at u0 ) 0.60 and 0.80 m s-1 elutriation rates of 0.25 and 1.6 g h-1, respectively, were measured. A further increase of u0 up to 1.0 m s-1 led to a more significant presence of solid in the exhaust gas with a corresponding elutriation rate of 4.50 g h-1, warning about the risk of not negligible catalyst loss when operating at very large gas velocities. 4. Conclusions Catalytic combustion of methane in a fluidized bed of a copper-based catalyst has been investigated in a bench-scale reactor with nonpremixed feeding of reactants. The reactor has been operated in the bubbling regime of fluidization, and very limited temperature differences inside the bed have been measured. Moreover, stable reactor operation has been observed during the entire experimental campaign, and bed activity reduction due to catalyst thermal deactivation or to catalyst loss by attrition did not occur to any significant extent. Conversion of methane was complete at temperatures ranging from 650 to 750 °C depending on the operative conditions in the reactor (inlet fuel concentration and fluidization velocity). As a consequence, under the conditions investigated, a maximum thermal power of about 4.5 kW could be produced by the reactor with fuel emissions lower than 100 ppm and without any formation of CO or NOx. Nomenclature Cin ) inlet methane concentration Cout ) outlet methane concentration Da ) Damko¨hler number
Ind. Eng. Chem. Res., Vol. 45, No. 3, 2006 1013
k ) apparent kinetic constant m ) reaction order with respect to fuel concentration Q ) total gas flow rate T ) temperature umf ) minimum fluidization velocity u0 ) superficial velocity Vcat ) volume of catalyst Literature Cited (1) Hayhurst, A. N.; John, J. J.; Wazacz, R. J. The Combustion of Propane and Air as Catalyzed by Platinum in a Fluidized Bed of Hot Sand. Proc. Combust. Inst. 1998, 27, 3111. (2) Simonov, A. D.; Yazykov, N. A.; Vedyakin, P. I.; Lavrov, G. A.; Parmon, V. N. Industrial Experience of Heat Supply by Catalytic Installations. Catal. Today 2000, 60, 139. (3) Mulder, A.; der Kinderen, J.; Riphagen, G. J. Catalytic Combustion: from Catalyst to Reactor. Proc. World Gas Conf. 1997, 20th, 485. (4) Marshall, K. J.; Mleczko, L. Catalytic Combustion of Hydrocarbons in an Internally Circulating Fluidized-Bed Reactor. Proc. Int. Conf. Circ. Fluid. Beds 1999, 6th, 539. (5) Foka, M.; Chaouki, J.; Guy, C.; Klvana, D. Natural Gas Combustion in a Catalytic Turbulent Fluidized Bed. Chem. Eng. Sci. 1994, 49, 4269. (6) Iamarino, M.; Chirone, R.; Lisi, L.; Pirone, R.; Salatino, P.; Russo, G. Cu/γ-Al2O3 Catalyst for the Combustion of Methane in a Fluidized Bed Reactor. Catal. Today 2002, 75, 317.
(7) Iamarino, M.; Chirone, R.; Pirone, R.; Salatino, P.; Russo, G. Catalytic Combustion of Methane and Propane in a Fluidized Bed Reactor. Proc. Combust. Inst. 2002, 29, 827. (8) Iamarino, M.; Chirone, R.; Pirone, R.; Salatino, P.; Russo, G. Catalytic Combustion of Methane in a Fluidized Bed Reactor under FuelLean Conditions. Combust. Sci. Technol. 2002, 174, 361. (9) Chyang, C. S.; Chang, C. H.; Chang, J. H. Gas Discharge Modes at a Single Horizontal Nozzle in a Two-Dimensional Fluidized Bed. Powder Technol. 1997, 90, 71. (10) Geldart, D. Types of Gas Fluidization. Powder Technol. 1973, 7, 285. (11) Sotudeh-Gharebaagh, R.; Chaouki, J.; Legros, R. Natural gas combustion in a turbulent fluidized bed of inert particles. Chem. Eng. Sci. 1999, 54, 2029. (12) Zukowski, W. The Role of Two-Stage Combustion in the Development of Oscillations during Fluidized Bed Combustion of Gases. Fuel 2000, 79, 1757. (13) Kunii, D.; Levenspiel, O. Fluidized Bed Reactor Models. 1. For Bubbling Beds of Fine, Intermediate and Large Particles. 2. For the Lean Phase: Freeboard and Fast Fluidization. Ind. Eng. Chem. Res. 1990, 29, 1226.
ReceiVed for reView September 10, 2005 ReVised manuscript receiVed November 29, 2005 Accepted November 30, 2005 IE051015E
