Experimental Investigation of Chemical-Looping Combustion in

Jan 18, 2011 - S. Noorman, F. Gallucci, M. van Sint Annaland,* and J. A. M. Kuipers. Chemical Process Intensification, Multiphase Reactors, Department...
0 downloads 0 Views 4MB Size
ARTICLE pubs.acs.org/IECR

Experimental Investigation of Chemical-Looping Combustion in Packed Beds: A Parametric Study S. Noorman, F. Gallucci, M. van Sint Annaland,* and J. A. M. Kuipers Chemical Process Intensification, Multiphase Reactors, Department of Chemical Engineering and Chemistry, Eindhoven University of Technology, Eindhoven, The Netherlands ABSTRACT: Recently a novel reactor concept for chemical-looping combustion (CLC) has been proposed exploiting dynamically operated packed beds. In this work, an experimental parametric study on this reactor concept has been carried out. The effects of operating temperature, feed oxygen concentration, steam addition, and type of oxygen carrier material (CuO/Al2O3 and CaMnO3) have been evaluated. It has been found that an increase of the reduction temperature resulted in a higher conversion of the oxygen carrier, increased reduction reaction rates, and increasing significance of various side-reactions. Steam addition suppresses carbon deposition on the oxygen carrier. Interesting insights in the relevance of steam reforming and water-gas shift during the reduction cycle were obtained. Finally, CaMnO3 is of interest for CLC in packed beds in view of the high oxygen capacity of the material and the excellent selectivity to the formation of CO2.

1. INTRODUCTION It is nowadays well accepted that carbon capture and sequestration (CCS), in which CO2 is recovered from (power plant) flue gas streams to be stored underground or mineralized, is one of the most promising short-medium-term technological options for reduction of anthropogenic CO2 emissions. For this reason, great effort is dedicated to make capture, transportation, and storage of carbon dioxide economically and technologically feasible for application in large-scale power production. Although different options for transportation and storage of CO2 are still under evaluation, it seems that CO2 capture is the most challenging part from an economic point of view. Conventional technologies for large-scale, low-emission power production are often classified into postcombustion, precombustion, and oxyfuel combustion processes.1 With precombustion and postcombustion capture additional CO2 separation units are necessary, while with oxy-fuel costly air separation units are required, leading to significant energy penalties. The success of CCS will be determined strongly by the extent to which these energy penalties can be minimized, either via the improvement of existing technologies (such as amine scrubbing) or via the development of novel technologies. One of the possibilities to inherently combine power production with the capture of pure CO2 in a single process step is chemical-looping combustion (CLC). The main advantage of CLC over the aforementioned options is that energy-intensive gas separation processes are avoided. In CLC, direct contact between air and fuel is circumvented, so that CO2 is obtained without nitrogen dilution. Instead, fuel and oxygen are contacted via an intermediate oxygen carrier, a metal/metal oxide that is alternately oxidized and reduced. The oxidation of the oxygen carrier is strongly exothermic, which is used to heat an air stream to very high temperatures. In principle, during the (often endothermic) regeneration with a fuel (in this work, natural gas is considered) only carbon dioxide and steam are formed, r 2011 American Chemical Society

which are easily separated via condensation. Additionally, with CLC the formation of thermal NOx is avoided due to the absence of the extremely high temperatures prevailing in a flame. CLC is commonly carried out by physically transporting the oxygen carrier particles between two interconnected fluidized beds, namely the air reactor (oxidation) and the fuel reactor (reduction).2 Alternatively, CLC using dynamically operated packed bed reactors was suggested and its technical feasibility was experimentally verified by Noorman and co-workers.3,4 In this concept, the oxygen carrier particles remain stationary and the gas flow is periodically switched. This process consists of alternate oxidation and reduction cycles, intermittently alternated with short periods of mild fluidization of the bed after each cycle to level off temperature and concentration profiles. The main incentive for performing the process in this way is that the difficult gas-solid separation needed in the fluidized bed approach is avoided, which is of vital importance considering that fines are produced that cannot be allowed to enter the downstream gas turbine. Moreover, the energy penalty for solid transportation is avoided. In this work, a detailed experimental parametric study for chemical looping combustion of methane in a dynamically operated packed bed has been carried out in order to elucidate the effects of temperature, oxygen carrier material, steam addition, and carbon deposition on the performance of the reactor.

2. EXPERIMENTAL PROCEDURE The chemical looping combustion has been carried out in a packed bed reactor. The experimental setup (schematically depicted in Figure 1) consists of a high-temperature-resistant Received: September 16, 2010 Accepted: December 23, 2010 Revised: December 7, 2010 Published: January 18, 2011 1968

dx.doi.org/10.1021/ie1019112 | Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

Figure 1. Schematic process flowsheet of the reactor setup. IR denotes the IR-analyzer.

Table 1. Experimental Conditions Used in This Work parameter

range

pressure, P

2.5 bar

oxidation temperature, Tox

450-800 C

reduction temperature, Tred inlet methane fraction (reduction), yCH4

500-800 C 4-10%

inlet oxygen fraction (oxidation), yO2

1-25%

stainless steel tube (OD  ID  L = 35  30  1500 mm), onto which 48 small tubes were welded to allow for the insertion of thermocouples (R€ossel, type K). The instationary axial temperature profiles (at the radial center) have been measured with rather good spatial resolution in the middle of the bed (temporal resolution was 1 Hz). To restrict the effect of radial heat losses via the reactor wall during operation the reactor was placed in a box filled with insulation material (vermiculite). Before starting an experiment, three electrical heating coils (600 W), which could be controlled separately, were used to establish a uniform initial axial temperature profile in the section where the active material was positioned (between 0.4 and 0.8 m): at the outer parts of the bed only inert material (alumina particles, dp = 3 mm) was used. A gas mixture consisting of 4-10% methane in nitrogen (total flow: 20 Ln 3 min-1) was fed to the reactor to reduce the oxygen carrier while simultaneously measuring the outlet concentrations of the relevant species with the IR analyzer. Next to the axial temperature profile, the outlet gas composition was monitored during the reduction cycles using an online IR/TCD-analyzer (Sick-Maihak s710), where the mole fractions of methane (0-5%), carbon dioxide (0-5%), and carbon monoxide (0-20%) were analyzed every 24 s, and the hydrogen fraction (0-40%) could be measured with very high temporal resolution. Dilution of the outlet stream with nitrogen could be applied to ensure that concentrations were within the operation ranges of the analyzer. The flows directed toward the reactor were controlled with automatic mass flow controllers (Brooks 5850s). Reactor operation and data acquisition were automated in LabVIEW (National Instruments). Two types of oxygen carrier (OC) materials were used in this work. The first OC was copper oxide on alumina, which was acquired from Sigma-Aldrich. The oxygen carrier was used as received from Sigma-Aldrich without pretreatments. The nominal composition of the particles was 13 wt % CuO/Al2O3. The second material tested was developed and provided by SINTEF, Norway. The particles used had a perovskite structure and consisted of a titanium-stabilized calcium manganese oxide (CaTi0.125Mn0.875O3, to be abbreviated in the following as CaMnO3). Particles with a 2-4 mm size, that were anticipated to be used in packed bed CLC, were available for testing. The TGA experiments used as benchmark in this work have been

Figure 2. Influence of the initial temperature during the oxidation cycle (gas flow rate =40 Ln 3 min-1): (a) maximum temperature rise as a function of the axial position in the bed and (b) local temperature rise as a function of time (axial positions x1 = 0.49 m, x2 = 0.59 m, and x3 = 0.69 m; empty symbols relate to Tox = 475 C, filled symbols relate to Tox = 710 C).

extensively reported in a recent paper.5 The experimental conditions investigated are reported in Table 1.

3. PARAMETER STUDIES Some of the main characteristics of the reactor concept, such as maximum temperature rise, temperature profiles, and front velocities, have been experimentally verified in a previous works.4,6,7 The influence of several relevant operating conditions is studied in this work. In the following, experiments in which copper oxide was used will be presented, investigating the effect of the initial temperature, the concentration of reactive species, and the possibility to add steam in order to suppress carbon deposition. After that, preliminary experiments performed with the CaMnO3 oxygen carrier will be discussed. 3.1. Effect of Oxidation Temperature. The influence of the initial uniform temperature at the start of the oxidation cycle on the (temporal and spatial) temperature evolution in the packed bed is shown in Figure 2. For the experiments shown, the preceding reduction cycles were carried out at an initial (uniform) temperature T0red of (approximately) 650 C, so that the maximally possible conversion of the material was comparable for these experiments. From Figure 2a, it can be discerned that the effect of the oxidation temperature on the maximum temperature rise in the packed bed is negligible, also considering the fluctuations in the mass change observed with thermogravimetry.5,8 In Figure 2b, it can be found that the rate at which the temperature increases at a fixed location (indicated in the figure 1969

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

Table 2. Conceivable Reactions during Oxidation and Reduction Cycles in Chemical-Looping Combustion over Copper Oxide on Alumina (Tref = 1073 K)a ΔHR(Tref) (kJ 3 mol-1)

reaction oxidations O2 þ 2Cu f 2CuO

-299.5

O2 þ C f CO2

-395.8

O2 þ 2C f 2CO

-224.8 reductions

CH4 þ 4CuO f 4Cu þ CO2 þ 2H2O CH4 þ CuO f Cu þ CO þ 2H2

-202.5 127.1

CO þ CuO f Cu þ CO2

-132.6

H2 þ CuO f Cu þ H2O

-98.5 gas phase reactions

CH4 þ H2O S CO þ 3H2

225.6

CO þ H2O S CO2 þ H2

-34.1

CH4 þ 2H2O S CO2 þ 4H2

191.5

CH4 þ CO2 S 2CO þ 2H2

259.8

carbon deposition CH4 f C þ 2H2 a

CO f 1/2 C þ 1/2 CO2

89.8 -170.0

Reaction heats are related to one mole of the first component mentioned in the reaction equation.

Figure 3. Outlet concentration profiles for a reduction cycle performed at (a) T0red = 650 C (open symbols denote measurements of the first experiment, closed symbols denote those of the second experiment) and (b) T0red = 800 C. At t = 0 s, a gas mixture consisting of 4% methane in nitrogen is fed to the reactor with previously oxidized oxygen carrier.

as x1 = 0.49 m, x2 = 0.59 m, and x3 = 0.69 m) does not change considerably as a function of the oxidation temperature either. Since the temperature rise is directly coupled to the reaction rate and the conversion of the oxygen carrier material, this also shows that the rate at which the material is oxidized is approximately equal for the selected cases. Moreover, when compared to the TGA experiments reported in ref 5 (not reproduced here), the time needed to fully oxidize the particles appears to be rather short (around 10 s here versus 25-30 s in the TGA experiment), but this can be explained considering that the pressure used in the experiments, and thus the partial pressure of oxygen, was higher than that with the TGA experiments (p = 2.5 bar versus atmospheric conditions). Furthermore, the evolution of the temperature obtained at position x3 suggests that the lower maximum temperature increase obtained with T0ox = 710 C is due to a decreased local conversion of the oxygen carrier material during the reduction cycle. It can be seen clearly that the temperature rise at this position is initially the same. This may have been caused by the thermal history of the material. It must be noted here that the experiments presented were not obtained from two consecutive cycles. It is concluded that there is no significant effect of the initial oxidation temperature on the observed reaction rate or on the temperature rise during the oxidation cycle. For this reason, in

the following, the oxidation temperature will no longer be mentioned explicitly in the experimental settings. However, it must be stressed that this temperature was chosen so that the maximum temperature in the bed would not exceed 1000 C, in order to avoid detrimental effects to the oxygen carrier and the reactor setup. 3.2. Effect of Reduction Temperature. Figure 3 shows the outlet concentration profiles for an experiment in which the initial reduction temperature was 650 C (Figure 3a) and 800 C (Figure 3b). When comparing the two experiments, a noticeable delay in the breakthrough of methane can be observed, indicating that the reduction rate has significantly increased with the increased temperature. Considering that the concentration of carbon dioxide already starts to decrease before the breakthrough of methane occurs, carbon is deposited onto the oxygen carrier during a large fraction of the reaction time. When the production of water (which was not detected) is ignored, the reaction stoichiometry of the reduction dictates that the concentration of carbon dioxide measured in the IR analyzer should be the same as the concentration of methane in the gas feed stream. As this is clearly not the case for the experiment shown here, it is concluded that methane is not only consumed via reaction with copper oxide, but also deposited onto the oxygen carrier surface, producing pure carbon and hydrogen. As, until the breakthrough of methane, no hydrogen is observed, hydrogen apparently reacts with the metal oxide available to form steam, and at a higher rate than with methane. It can be seen that the amount of carbon deposited is very relevant (Figure 3b): between t = 100 and 300 s, the amount of carbon formed is about 40% of the total amount of carbon added to the system in the form of methane in the feed stream. Finally, it is found that the deficit in the carbon balance decreases with increasing time, which indicates that the amount of carbon that can be deposited is limited. 1970

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

Figure 4. Local temperature evolution as a function of time for the experiment shown in Figure 3b. t = 0 s denotes the moment at which the feed stream was switched from inert to reducing gas. x0 (m) is the position in the bed relative to the boundary between the active and the inert material.

At the same time the breakthrough of methane occurs, significant production of carbon monoxide and hydrogen is observed. From the production of carbon monoxide it can be concluded that there must be some lattice oxygen left in the material, especially considering that the amount of CO measured decreased only slowly with t > 500 s. The production of carbon monoxide may originate from two mechanisms, which may both be of relevance: first, partial oxidation of the fuel may occur, producing carbon monoxide and hydrogen. From the reaction stoichiometry of this reaction, and the concentrations observed at t = 400-500 s, it is seen that this cannot be the only reaction of relevance. Alternatively, the hydrogen produced with the carbon deposition may react with oxygen in the oxygen carrier that is not so easily accessible, forming steam. Via steam reforming, now carbon monoxide may be produced. If only this reaction would occur, the ratio of the concentration of hydrogen and carbon monoxide should be 3:1, which seems to be close to the concentrations measured at t = 400-500 s. An overview of the conceivable main and side-reactions is given in Table 2. Further analysis on the relevance of these phenomena will be presented in Section 3.4. From the local temperature evolution depicted in Figure 4, it can be seen that the rate at which the temperature rises is lower than that with the oxidation cycle. This is not surprising considering that only a very low concentration of methane is used (4% CH4 in N2). Furthermore, two interesting observations can be made. First, in the very beginning of the reaction (around t = 10 s), there is a small temperature rise before the main temperature rise occurs. This may be caused by instationary effects. Second, it can be seen that the temperature decreases below the initial temperature, which is especially apparent at the beginning of the bed. This may be partially attributed to the endothermic deposition of carbon on the active material. Unfortunately, it is not possible to quantitatively address the relative importance of heat losses, the endothermic reaction, and the propagation of the heat front with the available model description.7 It must be noted that, as a consequence of the gas-solid reaction, a reaction front propagates through the bed. The

Figure 5. Influence of the initial temperature during the reduction cycle on (a) the temperature rise during the next oxidation cycle and (b) the local temperature rise during the oxidation cycle (x = 0.61 m).

reaction heat is of course released at the same speed of the reaction rate, and as a result of the reaction heat, the temperature of the bed changes, thus the reaction front has a temperature front associated with it. Along with the reaction front, another heat front propagates through the bed as a consequence of the temperature difference between the bed and the incoming gas flow. The rate at which this front moves is hereafter referred to as “heat front velocity”. From these observations it is clear that there is an important effect of the reduction temperature on the observed reaction rate, the amount of carbon formed, and the production of undesired byproduct. In Figure 5, the maximum temperature rise obtained during the oxidation cycles following the reduction cycles carried out at different temperatures is compared. An increase in the reduction temperature seems to lead to an increase in the capacity of the oxygen carrier and consequently to a larger maximum temperature rise during the oxidation cycle.5 From Figure 5a, it is found that this effect is considerable when the reduction temperature is increased from 500 to 650 C, while with T0red = 800 C, the additional temperature rise was found to be especially due to the combustion of carbonaceous species deposited on the material (while no carbon was deposited at lower temperatures as previously indicated in Figure 3). Finally, in Figure 5b, the effect of the initial temperature on the local temperature change is depicted. It can be seen clearly that the time needed to (apparently) completely convert the particle is similar in the three cases, although the maximum temperature rises and consequently the temperature gradients differ strongly. Differences in the particle morphology, affecting the relevance of 1971

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

intraparticle diffusion limitations, can be responsible for this effect. 3.3. Effect of Oxygen Concentration. In a previous paper3 it has been investigated whether, in a CLC process based on dynamically operated packed bed reactors, production of a constant high temperature gas stream during the oxidation cycle is in principle possible and how the operating parameters and properties of the oxygen carrier should be tuned to control the temperature of the outlet gas stream. In the following analysis an analytical expression for the maximum temperature is derived from an energy balance, where it is assumed that the packed bed initially consists of the completely reduced form of the oxygen carrier, which will react with the oxygen supplied in the gas feed stream with an infinitely high reaction rate. The exothermicity of this reaction causes the bed temperature to increase until complete particle conversion to the oxidized form has been attained. As a consequence of the convective gas flow, two heat fronts move through the reactor. In the analysis it is assumed that the heat capacity of the gas and solid and the solid density are constant and that the influence of the pressure drop and the variation of the mass flow rate, caused by the consumption of oxygen, can be neglected. Considering that the heat produced in the bed by the oxidation of the oxygen carrier is taken up by the solid material in the bed (neglecting the gas-phase volumetric heat capacity but not that of the solid phase), it has been demonstrated that the expected temperature rise in packed beds during CLC can be evaluated via3,4 ΔTmax ¼

ð - ΔHR , i Þ Cp, s Mact, j Cp, g Mg, i ωoact, j ςXj ωin g, i

ð1Þ

where Cp is the heat capacity of gas/solid phase in J kg-1 K-1, ΔTmax is the maximum temperature increase, ΔHR is the reaction enthalpy in J mol-1, M is the molecular weight in kg mol-1, ω is the weight fraction reactive material in solid or of gaseous components, Xj is the conversion of the solid material, and ζ is the stoichiometric factor. An interesting effect may occur when the two terms in the denominator in the equation are equal. Theoretically, this would mean that, under the assumption of infinitely fast reaction rates, the temperature rise obtained in the bed would be infinitely high. In Figure 6, this effect is demonstrated by varying the concentration of oxygen in the feed stream during the oxidation cycle. Again, all reduction cycles were executed at the same initial temperature, so that the average conversion over the bed was similar for the experiments. It can be seen that there is indeed an influence of the oxygen content in the gas feed stream on the temperature rise, where decreasing the oxygen concentration in the feed results in a significant increase in the temperature rise, but this effect is not as large as might have been expected. The higher temperature rise in case of a lower oxygen concentration in the feed is related to the increased energy accumulation at the reaction front, because the heat front velocity (proportional to the relative volumetric heat capacity) approaches the reaction front velocity (proportional to the oxygen supply rate). With higher oxygen concentrations in the feed, the heat front velocity is much slower than the reaction front velocity and a temperature plateau is created. These are the conditions of interest for energy production. Based on the physical properties of the oxygen carrier used, the singularity predicted in eq 1 would be expected at approximately

Figure 6. Influence of varying the oxygen content in the gas feed stream on the local temperature evolution (data were collected in the middle of the bed) in packed bed CLC experiments. All preceding reduction cycles were carried out at T0red = 650 C.

yO2 = 0.025. In the current case however, heat losses, axial conduction, and a finite reaction rate limit the temperature rise. 3.4. Steam Addition and Carbon Deposition. As has been demonstrated in the previous sections, with the oxygen carrier used, there is an important effect of carbon deposition both during the reduction and the oxidation cycles. Carbon deposition, which should be avoided to maximize the CO2 capture efficiency and to avoid material degradation, may be suppressed by recycling the flue gas stream or by addition of pure steam. In this section, the effect of steam addition will be discussed. In the reduction experiments, between 0 and 5 mL 3 min-1 (liquid) water was fed via an HPLC pump (Gilson 302), while the inlet flow (steam excluded) was kept constant at 20 Ln 3 min-1, with 5% of methane. To be able to quantitatively estimate the amount of carbon deposited, oxidation cycles were carried out using a 20 Ln 3 min-1 air flow that was diluted with nitrogen so that yinO2 = 0.10. During the experiments (both reduction and oxidation), no conversions could be observed when only steam was added, so that water splitting, which is known to occur with iron oxide, can be ruled out. The experiments were carried out at initial reduction temperatures of 700 and 800 C. It must be realized that with the addition of steam, a large range of additional reactions are conceivable that may be catalyzed by the oxygen carrier, either in its oxidized or reduced form. In Table 2, an overview of possible reactions in chemicallooping combustion and the associated reaction heats was given. Of special interest for this section are the reforming and shift reactions. The aim of steam addition is to suppress carbon deposition, which can be formed either via decomposition of methane or via the Boudouard reaction. In Figure 7a, the effect of steam addition (in this experiment Φl = 3 mLl 3 s-1, corresponding to a methane-to-steam ratio of approximately 1:4) on a reduction cycle performed with an initial temperature T0red = 800 C is depicted. When these concentration profiles are compared with those shown in Figure 3b, some significant differences can be observed. First, it is seen that the desired production of pure carbon dioxide can indeed be achieved with steam addition, and from the fact that the concentration of CO2 remains constant for some time, carbon deposition can be ruled out. After the breakthrough of methane, the formation of the byproduct hydrogen and carbon monoxide is observed in much higher concentrations than when methane was only diluted with nitrogen. This indicates that there is an increased relevance of additional (gas-phase or catalytic) 1972

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

Figure 7. Effect of steam addition on concentration profiles at the reactor exit for (a) a reduction cycle executed at T0red = 800 C and (b) oxidation cycles with and without steam addition. In the steam addition experiments, yinCH4 = 0.05 (on dry basis), and the methane-to-steam ratio in the gas feed mixture was approximately 1:4.

Figure 8. Local temperature rise as a function of time for experiments with (dashed lines) and without (solid lines) steam addition (Φl = 3 mL 3 s-1): (a) reduction cycle, (b) oxidation cycle.

reactions. The effect of steam addition on carbon deposition has been (indirectly) demonstrated in Figure 7b as well, where the production of carbon dioxide in the oxidation cycle is shown. Clearly, carbon deposition has been almost completely suppressed. In Figure 8, the effect of steam addition on the local temperature evolution during oxidation and reduction cycles is depicted. For the oxidation cycle (Figure 8b), the maximum temperature rise is found to decrease when steam is added, which can be explained considering that the contribution of the combustion of carbon can now be neglected. The temperature rise found suggests that the conversion of the active material is close to 100%, based on an active weight content of 12.5%4). Moreover, as the amount of oxygen needed to completely convert a certain volume of bed material has decreased due to the absence of carbon on the oxygen carrier, the reaction front velocity increased when steam was added. With the reduction cycle, a surprising effect can be seen in Figure 8a. It is observed that the addition of steam results in a larger maximum temperature rise than when methane was only diluted with nitrogen. Moreover, in both cases the temperature increase observed is larger than that anticipated based on the complete conversion of copper oxide with methane (see ref 4), albeit only slightly with the latter. This is interesting, since both methane pyrolysis and steam reforming, which are considered to be the main additional reactions for this system, are endothermic

processes, so that temperature changes would actually be expected to be smaller than when only the reduction of the oxygen carrier would be taken into account. For the case in which carbon deposition is of importance, this can be explained by the fact that due to this reaction, the concentration of methane in the bed decreases. The consequence of this is that the phenomenon discussed in the previous section may occur: due to a lower weight fraction of methane in the bed, the reaction front velocity decreases and consequently the maximum temperature rise in the bed increases. It can be expected that when more realistic gas feed conditions would be used (i.e., a larger methane content in the gas feed stream), this effect would be of less importance, especially when, as was suggested in the previous section, the maximum amount of carbon that can be deposited on an oxygen carrier particle is limited. When steam addition is applied, steam reforming and watergas shift may occur, especially when the oxygen carrier is reduced. Although the shift reaction is exothermic, the net heat effect of these conversions will be endothermic. When the zone in the bed that is reduced increases, more reforming will occur, thus decreasing the temperature in the bed and effectively increasing the “heat front velocity”. As the reaction front velocity remains the same, the difference between the heat and reaction front velocities decreases during the reduction cycle, so that the temperature rise in the bed increases with the propagation of the fronts. To further clarify these effects, temperature profiles 1973

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

Figure 9. Effect of the steam content in the gas feed mixture (yinCH4= 0.05). The production of CO and CO2 during (a) the reduction cycle (carbon monoxide production at T0red= 700 C was negligible for all cases) and (b) the oxidation cycle. The total amount of methane fed was approximately the same for all the experiments.

obtained from numerical simulations for three relevant cases of reduction are presented in Appendix A. Finally, in Figure 9, the effect of the steam content in the gas mixture is investigated. In the first place, it is found that with an initial reduction temperature of 700 C, the effect of steam addition is rather small. As a consequence of this, the production of CO2 is more or less constant and, as can be seen in Figure 10a, only a rather small effect is found on the maximum temperature rise obtained during either the oxidation or the reduction cycle. This shows that at this temperature, neither the effect of carbon deposition nor that of side-reactions is of much importance. For the experiments performed at an initial temperature of 800 C, more significant effects can be discerned. In Figure 9b, the amount of carbon dioxide produced during oxidation cycles, originating from the combustion of solid carbon, is shown. It is concluded that the steam content needed to completely suppress carbon deposition is somewhat higher than the 1:1 ratio (yinH2O = 0.05) expected based on thermodynamic calculations.9 The effect of steam reforming and shift reactions starts to become apparent only when the effect of carbon deposition is almost negligible. Then, the production of CO and CO2 increases, indicating an increased relevance of catalytic and/or gas phase reactions, which leads to an increase in the maximum temperature rise obtained during the reduction cycles and, as carbon deposition is prevented, in a decreased maximum temperature rise with the oxidation cycles (see Figure 10a). An additional experiment was performed in which the methane content in the

ARTICLE

Figure 10. Effect of the steam content in the gas feed mixture (yinCH4= 0.05): (a) the maximum temperature rise obtained during the reduction and subsequent oxidation cycles, and (b) averaged concentrations found when the complete bed is reduced (at T0red = 800 C). It must be noted that in the experiment performed at 800 C and yinH2O = 0.25, the methane content in the gas feed was yinCH4 = 0.10. The total amount of methane fed was approximately the same for all the experiments.

gas feed mixture was increased with a factor two (the total flow rate was kept the same), and showed the same trends. However, a significantly lower temperature rise is obtained during the reduction cycle, while that obtained with the oxidation cycle was the same. This can be explained by the fact that when a higher methane content is used, the reaction front velocity increases, while the heat front velocity remains the same. Thus, as the difference in the front velocities has increased, the effect of steam reforming as discussed before is now of less importance. In Figure 10b the outlet concentrations at the moment when the complete bed was reduced are shown (cf. Figure 7a, t = 350-450 s). Although these are not totally constant, due to variations in the bed temperature, the average values obtained give an indication of the catalytic activity of the reduced oxygen carrier for steam reforming and shift processes. At low steam concentrations, only the production of hydrogen, resulting from carbon deposition, and the breakthrough of unconverted methane could be observed, whereas when steam was added, the production of both CO and CO2 became significant. 3.5. Application of CaMnO3 in Packed Bed CLC. Titaniumstabilized CaMnO3 has been found to be an interesting oxygen carrier candidate with a high oxygen capacity,10 a good resistance against carbon deposition, and, unfortunately, low stability. Detailed TGA experiments have been performed also on this material.9 In this section, a number of experiments performed with this material will be presented, so that improvements to the material can be suggested based also on operation in the packed bed reactor system. 1974

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

For the experiments, the same setup as with the copper oxide oxygen carrier was used. The reactor was filled with 150 g of CaMnO3, with a particle size of 2-4 mm and a rather low bulk density of around 590 kg 3 m-3. When the reactor was emptied, serious attrition was observed, meaning that the mechanical stability of the material needed to be improved. The gas feed during the reduction and oxidation cycle was 20 and 40 L 3 min-1, respectively. Experimental results obtained with this system are presented in Figures 11-13. First, in Figure 11a, the concentration of carbon dioxide measured with the IR analyzer is shown for experiments at different reduction temperatures and methane content in the gas feed stream. It must be noted that for none of the experiments performed with this material the production of hydrogen, carbon monoxide or, during the oxidation cycle, solid carbon combustion products could be discerned, indicating that the selectivity of the material is very favorable for CLC. For this reason, in the figure only the concentration profile of carbon dioxide is shown; the breakthrough curve of methane follows directly from the mass balance. Additionally, during the reduction cycle, hardly any temperature change could be observed. In case of the experiment performed at T0red = 800 C, a maximum temperature rise of approximately 30 C was obtained. When this is compared to the heat effect during the oxidation cycle, it can be concluded that the reaction heat for the reduction of the oxygen carrier is very small. The possible reactions involved during oxidation and reduction of this oxygen carrier are 4CaMnO3 þ CH4 f 4CaMnO2 þ CO2 þ 2H2 O 2CaMnO2 þ O2 f 2CaMnO3

ðreductionÞ

ðoxidationÞ

However a more specific study on the reactions occurring during CLC is required since it is known that CaMnO3 can be also reduced to CaMnO2,5.14-16 An important disadvantage of this material is that the oxygen carrier displays only a rather low reactivity at the temperatures of interest. This can be observed from the immediate breakthrough of methane. From the TGA experiments with CaMnO3 it can be concluded that when higher temperatures were applied in the experiments, the apparent reaction rate would be significantly higher.9 However, to avoid even larger temperature rises during the oxidation cycle than already observed (which could be detrimental to both the oxygen carrier material itself and to the reactor setup), no reduction cycles were performed at higher temperatures than T0red = 800 C. It must be noted that also in an industrial application it is unlikely that the reduction cycle would be carried out at very high temperatures, since in this case a lowoxygen-capacity oxygen carrier would need to be used to arrive at suitable maximum temperatures during the oxidation cycle. The low reaction rate observed with the reduction cycle is also (at least partially) responsible for the lack of uniformity in the maximum temperature rise found in the oxidation cycles. In all of the experiments, a gradient in the maximum temperature rise is found (see Figure 11b), from which it is concluded that nonuniform conversion of the oxygen carrier was achieved. It must be noted that in the first few reduction-oxidation cycles this gradient was even larger. From the maximum temperature rise obtained in the oxidation cycle it is found that, compared to the CuO/Al2O3 oxygen carrier, the capacity of this material is very large, as was also anticipated from a TGA study.9

Figure 11. (a) Reduction and (b) oxidation cycle characteristics of the CaMnO3 oxygen carrier in packed bed chemical-looping combustion. Oxidation cycles were carried out at rather low temperature (T0ox < 400 C) to avoid excessively high temperatures in the reactor.

Based on the experience with copper oxide and the results obtained with the TGA study (not reported here), the effect of the oxidation temperature on the reaction characteristics was expected to be negligible. This assumption was supported by the very high reaction rates observed even when the oxidation was carried out at an initial temperature of only 300 C, indicating that the ignition temperature for the reaction is very low. Additionally, in Figure 12, local temperature evolutions during an oxidation cycle are shown. For this specific case, a very high maximum temperature rise of up to 800 degrees is observed. At its maximum, the rate at which the temperature rises is found to be close to 120 K 3 s-1. As a consequence of the lower conversion realized further in the bed, both the observed maximum temperature rise and the reaction rates become lower with the propagation of the reaction front through the bed. Obviously, this results in an increase of the reaction front velocity with the advancing of the fronts as well. To be able to compare the results obtained in the packed bed chemical-looping combustion experiments with those obtained with TGA, an estimate of the oxygen capacity of the material must be obtained. In the packed bed experiments, it is possible to obtain the capacity both from the reduction and the oxygen carrier cycle experiments. The average capacity of the oxygen carrier material, which is represented by the amount of oxygen consumed in the material δ was estimated from the amount of CO2 produced in the reduction cycle 4nCO2, prod ð2Þ Æδæred ¼ 0 nCaMnO3 where n0CaMnO3 is estimated from the mass of the oxygen carrier in the system and the molecular weight of the material. 1975

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

Figure 12. Local temperature evolution and heat production rate for an oxidation cycle with the preceding reduction cycle carried out at T0red = 800 C. Coordinates x correspond to the axial positions found in (b).

For the oxidation cycle, the capacity of the material could be estimated from the reaction front velocity. The following description of δ could be obtained9 (with ωact,ox = 1.0, since no support material was used): Æδæox ¼

2Mact, ox Mo2 1 þ

Fbulk Fg νg ωin o2

!

ð3Þ

In Figure 13, a comparison between the capacity of the material as predicted from the reduction and the oxidation cycles is given. It is found that there is a good correspondence between these methods, especially considering that, due to the variation of the conversion of the bed, the temperature rise and reaction front velocity are not uniform over the entire bed and that potential variations in the physical properties of the oxygen carrier with temperature (e.g., heat capacity, density) were neglected. This suggests that the predicted oxygen capacity is reliable. It must be noted that in this comparison, only the last ten cycles performed with this material were considered. In the earlier experiments, the lack of uniformity in the temperature profile for the oxidation cycle made a proper estimation of the front velocity impossible. As expected, in Figure 13b it can also be seen that when the reduced state of the oxygen carrier contained less oxygen, the average maximum temperature rise obtained over the bed was strongly increased. When the capacity of CaMnO3 as derived in the packed bed CLC experiments is compared to that obtained with the TGA experiments, it can be concluded that the maximum capacity of

Figure 13. Oxygen capacity of CaMnO3 in packed bed chemicallooping combustion experiments: (a) correspondence between the average δ estimated from the reduction and the oxidation cycles (labels indicate the average initial reduction temperature T0red) and (b) the effect of the capacity on the average maximum temperature rise during the oxidation cycle.

the oxygen carrier has not been achieved. The reason for this is unclear: when the reaction was stopped, the concentration of CO2 measured was close to zero in all experiments, suggesting that the maximum conversion of the material was actually obtained. Perhaps, a decreased activity due to high temperature stresses encountered caused lower conversions. Finally, it can be concluded that in the current form, the material used is not suitable for application in packed bed chemical-looping combustion. First, the stability of the material was not guaranteed over a large number of cycles, as was also found in the TGA study. Moreover, the oxygen carrier exhibits only a high reactivity during the reduction cycles with rather high temperatures (1000 C), so that considerable losses due to methane slip or the inefficient use of oxygen carrier materials can be anticipated when lower temperatures are applied.17 To improve stability and reactivity and to limit the temperature rises obtained with the oxidation cycles, it may be considered to combine the use of CaMnO3 with a support material or simply dilute the system with an inert (or another oxygen carrier with higher reactivity for the reduction). Considering the excellent selectivity of the material during reduction cycles and the high reactivity with the oxidation cycle the material remains of interest for application in packed bed chemical-looping combustion, but requires further development. 1976

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

4. CONCLUSIONS In this paper a parametric study of CLC process in packed bed has been carried out. The analysis was related to experiments performed using the CuO/Al2O3 oxygen carrier already studied in a previous work. In accordance with the TGA experiments, an important effect of the reduction temperature on the reaction characteristics was observed and an increase of the reduction temperature resulted in a higher conversion of the oxygen carrier, increased reduction reaction rates and increasing significance of various side-reactions. From the addition of steam, interesting insights into the relevance of carbon deposition, steam reforming, and water-gas shift during the reduction cycle were obtained. Most importantly, it was found that with the addition of steam, carbon deposition could be suppressed completely with a steam-to-methane ratio in the gas feed stream slightly higher than one. The actual application of copper oxide in an industrial-scale chemical-looping combustion process remains unlikely in view of the lower process efficiency that can be achieved with the use of a relatively low reaction temperature, which is necessary considering the low melting point of the material. The use of a higher temperature could lead to serious structural changes inside the particle, endangering the long-term stability and predictable behavior of the oxygen carrier, which, too, are vital for the success of the application as a whole. CaMnO3 is of interest in view of the high oxygen capacity of the material and the excellent selectivity to the formation of CO2. However, in order to solve problems related to the low stability of the material and the low reaction rates observed during the reduction cycle, several improvements to the material are required before it can become an interesting candidate for industrial application, especially in view of the significant material costs. ’ APPENDIX A. INTERPRETATION OF THE STEAM ADDITION To improve the understanding of the effect of steam addition on the temperature evolution during the reduction cycle (see Figure 8a), results from numerical experiments using the model description discussed in ref 3 are presented in Figure A1. In the graphs, the evolution of the axial temperature profile in packed bed chemical-looping combustion is shown for three different scenarios for the reduction cycle: (1) reduction with methane, (2) reduction and carbon deposition, (3) reduction and steam reforming. In the simulations, the bed was assumed to be filled with a 10 wt % CuO/Al2O3 oxygen carrier that was initially fully oxidized and had a length of 50 cm, enclosed by two small inert zones. The gas feed stream (balance N2) consisted of 5 vol % methane and an additional 15 vol % of steam for the third case (Figure A1c). Heat losses were not taken into account. The reduction of the oxygen carrier with methane, hydrogen, or carbon monoxide was described using first-order reactions and effective rate constants (Table A1), as was the case with carbon deposition via methane decomposition. It was assumed that carbon deposition could occur only when the oxygen carrier was completely reduced. In case of steam addition, steam reforming and water-gas shift reactions were assumed only to take place on a reduced surface. For these reactions, (temperature dependent) gas phase equilibria were used and effective rate

ARTICLE

Table A1. Effective Rate Constants Used for the Simulations keff

reaction reductions: ri = keffCg,i CH4 þ 4CuO f 4Cu þ CO2 þ 2H2O

5.0

CO þ CuO f Cu þ CO2

10.0

H2 þ CuO f Cu þ H2O

5.0

gas phase reactions: ri = keffCg,1Cg,2 - (keff/K)Cg,3Cg,4 CH4 þ H2O S CO þ 3H2

5.0

CO þ H2O S CO2 þ H2

5.0

carbon deposition: ri = keffCg,i CH4 f C þ 2H2

2.0

CO f 1/2 C þ 1/2 CO2

5.0

constants were chosen so that the concentrations of the species at the reactor outlet were of the same order of magnitude as encountered with the experimental results. When the reduction with H2 or CO was of relevance (case 2 and 3), these reactions were assumed to proceed at a higher rate than with methane, so that methane would break through first (as was the case with the experiments). In Figure A1a, it is found that when only the reduction with methane is included, a temperature profile as predicted in ref 3 is found. However, when additional reactions such as carbon deposition (see Figure A1b) are of relevance, these predictions no longer hold. As a consequence of carbon deposition in the fully reduced part of the packed bed, the concentration of methane decreases over the reactor length. As a consequence of this, the reaction front velocity is smaller than when only reduction of methane would be included, and it also decreases as the zone in which carbon deposition may occur increases. As can be found from eq 1, the temperature rise is inversely proportional to the difference of the front velocities. Thus, the temperature rise will increase with the axial position in the bed, until only hydrogen would be available for the reduction of the metal oxide. What would happen in the case of higher concentrations of methane strongly depends on reaction orders and temperature dependency in the reaction kinetics of the different reactions. When steam addition is applied, it can be seen that the reaction front velocity is the same as with the reduction, with methane in the absence of mass deposition on the fully reduced oxygen carrier. Still, the maximum temperature rise in the bed can be found to increase with the axial position in the bed. In this case, due to the endothermic steam reforming, effectively the heat front velocity is increased (although in this case it cannot really be referred to as a heat front). The mechanism now is the same as with carbon deposition: as the difference between the front velocities decreases when the effect of steam reforming increases, the temperature rise in the bed increases while the fronts propagate. It must be noted that as the steam reforming and carbon deposition kinetics are unknown the profiles shown only have a qualitative value. Moreover, in the simulations shown, mass transfer limitations were ignored. Still, it is believed that these profiles give a reasonable description of what may occur in the experiments and properly indicate the effect of steam addition and carbon deposition in packed bed reactor experiments for chemical-looping combustion. 1977

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

Table B1. Governing Equations, Boundary Conditions, and Dimensionless Numbers for the Description of Combined Reaction and Heat Losses in a Cylindrical System (See Also Ref 10) energy balance:

dimensionless:

boundary conditions:

Fbulk Cp, bulk ∂T ∂t ¼

1 ∂ ∂T r ∂r rλrad, ef f ∂r þ q

Fbulk Cp, bulk ∂T ∂t ¼

1 ∂ ∂T r ∂r rλrad, ef f ∂r

∂Yðξ, FoÞ ∂Fo

¼

1 ∂ ∂Y ξ ∂ξ ξ ∂ξ

∂Yðξ, FoÞ ∂Fo

¼

1 ∂ ∂Y ξ ∂ξ ξ ∂ξ

r ¼ 0,

∂T ∂r

¼ 0 f ξ ¼ 0,

þ Fo1 0

for t < τr

for t > τr

for Fo < Fo0

for Fo > Fo0 ∂Y ∂ξ

¼ 0

r ¼ R, λrad;eff ∂T ∂r ¼ RðT - T0 Þ f ξ ¼ 1, ∂Y ∂ξ

¼ BiY

t ¼ 0, T ¼ T0 f Fo ¼ 0, Y ¼ 0 λ

rad;eff r dimensionless numbers: Fo ¼ Rat2 , Fo0 ¼ aτ R 2 , a ¼ Fbulk Cp, bulk , ξ ¼ Tðr, tÞ - T¥ RR Tad , Bi ¼ λrad;eff , ∂Y ðξ, FoÞ ¼ Δ

ΔTad ¼

ωact ð - ΔHR Þ Cp, s Mact ,

q ¼

r R0 ,

Fbulk ωact ð - ΔHR Þ Mact τr

Table B2. Variables Used for the Estimation of Radial Temperature Profiles in Packed Bed CLC parameter

Figure A1. Qualitative description of the temperature evolution (t = 1, 2, 3, 4, 5 min) obtained from numerical simulations for several reaction cases: (a) reduction with methane, (b) reduction with methane and carbon deposition, and (c) reduction with methane and steam reforming. Kinetic parameters used can be found in Table A1.

’ APPENDIX B. EVALUATION OF HEAT LOSSES To evaluate the relevance of radial heat losses during experiments in the packed bed reactor system, the temperature profile in a cross-section of this reactor is calculated. In the current analysis (see Table B1 for the governing equations and the boundary conditions supplemented), the following representation of the actual system is used: at t = 0, both the packed bed and its surroundings (the reactor wall and the insulation material) will be at an initial temperature T0. Now it is assumed that due to reaction, heat will be produced at a constant rate q, for the total reaction time τr. As only the investigation of radial

value

unit

R

0.015

m

Fbulk

1000

kg m-3

Cp,s

1200

J kg-1 K-1

ΔHR

-300

kJ mol-1

λrad,eff

2.0

W m-1 K-1

R

300

W m-2 K-1

Bi

2.25

-

τr Fo0

15 0.11

s -

heat losses is of importance here, heat dispersion in the axial direction is neglected in this scenario, so that the expected maximum temperature rise in the system equals the adiabatic temperature rise. For the heat losses, a worst case scenario is considered by assuming that the surroundings of the reactor act as a heat sink and will remain at the initial temperature T0. When the reaction is completed (t > τr), only heat losses occur. A description of the radial temperature profile inside the packed bed reactor in case both heat production and heat transfer to the surroundings occur can be obtained by solving the partial differential equations presented in Table B1 (see Carslaw and Jaeger11), yielding   1 2 2 1 ξ þ Y ðξ, FoÞ ¼ 4Fo0 Bi ¥ 2Bi X J0 ðξβn Þ exp½ - Foβ2n  2 2 - 0 ðB.IÞ Fo n ¼ 1 βn ðBi þ β2n ÞJ0 ðβn Þ for Fo < Fo0 (note that q and τr are correlated) and where βn are the roots of the following characteristic equation: βJ1 ðβÞ - BiJ0 ðβÞ ¼ 0

ðB.IIÞ

When reaction is complete and only heat losses occur (in this case, the initial temperature profile corresponds to the 1978

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

The values used in the calculations (see Table B2) were obtained from the wall-to-bed heat transfer correlations by Dixon and Creswell12 (see also Smit13). From Figure B1 it is concluded that due to the rather short Fourier times encountered, even though the effect of heat losses can be very relevant at the outer part of the bed, the temperature rise in the center of the bed is hardly affected, which is where in the experiments the temperature was measured. The temperature in the center of the bed is more than 98% of the maximum temperature rise in cases where the systems were adiabatic. For the case presented, the heat losses are probably even overestimated: in reality the reactor wall (and insulation material) will heat up due to the heat losses, thus decreasing the driving force for the heat losses. In Figure B2, it is shown that if the total reaction time is increased, the effect of heat losses obviously increases significantly. It is concluded that for the oxidation cycle experiments, in which the reaction times are rather short, the observed temperature rise in the packed bed reactor is hardly influenced by heat losses. A more exact description of the effect of heat losses is quite complicated, considering the highly dynamic nature of the experiments performed, both inside the reactor and in the surroundings.

’ AUTHOR INFORMATION Figure B1. Estimated dimensionless radial temperature profiles for different Fourier times for two cases: (a) chemical reaction and heat losses, and (b) only heat losses (for the settings used, see Table B2).

Figure B2. Dimensionless temperature at different radial coordinates as a function of the total reaction time τr at Fo = Fo0 . Parameter values are listed in Table B2.

profile obtained from eq B.I at Fo = Fo0 ), the following equation applies: Y ðξ, FoÞ ¼

Z 1 β2 J0 ðξβ Þ 2 exp½- Fo2 β2n  2 n 2 n ðBi þ βn ÞJ0 ðβn Þ 0 n¼1 ¥ X

ξY ðξ, Fo

0

ÞJ0 ðξβn Þdξ ðB.IIIÞ

where Fo2 = Fo - Fo0 . With eqs B.I- B.III, the evolution of the radial temperature profiles can be calculated (the integral in eq B.III was evaluated numerically and the series were calculated for n = 1-10). The physical properties and experimental conditions given in Table B2 were used for these calculations. Especially the estimation of the effective radial heat conductivity λrad,eff and the overall heat transfer coefficient R deserves attention, as there are a large number of correlations available, not all yielding similar predictions.

Corresponding Author

*E-mail: [email protected]. Tel.: þ31 40 247 2241. Fax: þ31 40 247 5833.

’ NOMENCLATURE a = effective thermal heat conductivity (m2 3 s-1) Bi = Biot number, Bi = RR/λrad,eff (-) Cp = heat capacity of gas/solid phase (J 3 kg-1 3 K-1) Fo = Fourier number, Fo = at/R2 (-) Fo0 = Fourier number, Fo0 = aτr/R2 (-) ΔHR = reaction enthalpy (J 3 mol-1) L = reactor length (m) Mi = molecular weight of component i (kg 3 mol-1) p = pressure (Pa) r = radial position (m) t = time (s) T = temperature (K) ΔTmax = maximum temperature rise (K) vg = gas velocity (m 3 s-1) w1 = reaction front velocity (m 3 s-1) w2 = heat front velocity (m 3 s-1) x = axial position (m) X = conversion (-) Y = dimensionless temperature (-) YL = dimensionless temperature at reactor outlet (x = L) (-) ’ GREEK LETTERS ε = porosity (-) ζ = stoichiometric factor (-) λeff = effective heat dispersion (W 3 m-1 3 K-1) F = density (kg 3 m-3) τr = L/w1 = reaction front time (s) ξ = dimensionless radial position (-) ωact = weight fraction of reactive material in solid (-) ωg,i, ωs,j = weight fraction of gas/solid material (-) 1979

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980

Industrial & Engineering Chemistry Research

ARTICLE

’ SUB- AND SUPERSCRIPTS bulk = bulk conditions g = gas phase i, j = component in the gas/solid phase in = inlet condition ox = oxidative state red = reductive state s = solid phase ’ REFERENCES (1) IPCC. IPCC Special Report on Carbon Dioxide Capture and Storage; Metz, B., Davidson, O., de Coninck, H. C., Loos, M., Meyer, L. A., Eds.; Prepared by Working Group III of the Intergovernmental Panel on Climate Change; Cambridge University Press: Cambridge, UK and New York, 2005. (2) Lyngfelt, A.; Leckner, B.; Mattisson, T. A fluidized-bed combustion process with inherent CO2 separation; application of chemicallooping combustion. Chem. Eng. Sci. 2001, 56, 3101–3113. (3) Noorman, S.; van Sint Annaland, M.; Kuipers, J. A. M. Packed bed reactor technology for chemical-looping combustion. Ind. Eng. Chem. Res. 2007, 46, 4212–4220. (4) Noorman, S.; van Sint Annaland, M.; Kuipers, J. A. M. Experimental validation of packed bed chemical-looping combustion. Chem. Eng. Sci. 2010, 65 (1), 92–97. (5) Noorman, S.; Gallucci, F.; van Sint Annaland, M.; Kuipers, J. A. M. An experimental investigation of a CuO/Al2O3 oxygen carrier for chemicallooping combustion. Ind. Eng. Chem. Res. 2010, 49 (20), 9720–9728. (6) Noorman, S.; Gallucci, F.; van Sint Annaland, M.; Kuipers, J. A. M. A theoretical investigation of CLC in packed beds. Part 1: particle model. Chem. Eng. J. 2010, in press, doi:10.1016/j.cej.2010.12.068. (7) Noorman, S.; Gallucci, F.; van Sint Annaland, M.; Kuipers, J. A. M. A theoretical investigation of CLC in packed beds. Part 2: reactor model. Chem. Eng. J. 2010, in press, doi:10.1016/j.cej.2011.01.012. (8) Garcıa-Labiano, F.; Adanez, J.; de Diego, L. F.; Gayan, P.; Abad, A. Effect of pressure on the behavior of copper-, iron-, and nickel-based oxygen carriers for chemical-looping combustion. Energy Fuels 2006, 20 (1), 26–33. (9) Noorman, S. Packed Bed Reactor Technology for Chemicallooping Combustion. PhD thesis, University of Twente, Enschede, the Netherlands, 2009. (10) Leion, H.; Larring, Y.; Bakken, E.; Bredesen, R.; Mattisson, T.; Lyngfelt, A. Use of CaMn0.875Ti0.125O3 as oxygen carrier in chemicallooping with oxygen uncoupling. Energy Fuels 2009, 23 (10), 5276– 5283. (11) Carslaw, H. S., Jaeger, J. C. Conduction of Heat in Solids; Oxford University Press Inc.: New York, 1959. (12) Dixon, A. G.; Creswell, D. L. Theoretical prediction of effective heat-transfer parameters in packed-beds. AIChE J. 1979, 25 (4), 663. (13) Smit, J. Reverse Flow Catalytic Membrane Reactors for Efficient Syngas Production. PhD thesis, University of Twente, Enschede, the Netherlands, 2006. (14) Rørmark, L.; Mørch, A. B.; Wiik, K.; Stølen, S.; Grande, T. Enthalpies of oxidation of CaMnO3-δ, Ca2MnO4-δ and SrMnO3-δ Deduced redox properties. Chem. Mater. 2001, 13 (11), 4005–4013. (15) Poeppelmeier, K. R.; Leonowicz, M. E.; Scanlon, J. C.; Longo, J. M.; Yelon, W. B. Structure determination of CaMnO3 and CaMnO2.5 by X-ray and neutron methods. J. Solid State Chem. 1982, 45 (1), 71–79. (16) Poeppelmeier, K. R.; Leonowicz, M. E.; Longo, J. M. CaMnO2.5 and Ca2MnO3.5: New oxygen-defect perovskite-type oxides. J. Solid State Chem. 1982, 44 (1), 89–98. (17) Fossdal, A.; Bakken, E.; Øye, B. A.; Schøning, C.; Kaus, I.; Mokkelbost, T.; Larring, Y. Study of inexpensive oxygen carriers for chemical looping combustion. Int. J. Greenhouse Gas Control 2010 Article in Press.

1980

dx.doi.org/10.1021/ie1019112 |Ind. Eng. Chem. Res. 2011, 50, 1968–1980