Chemical Looping Combustion of Solid Fuels in a 10 kWth Unit

Chemical Looping Combustion of Solid Fuels in a 10 kWth Unit — The present study is based on previous results from batch experiments which were conducted in a 10 kWth chemical looping combustor for solid fuels using ilmenite, an iron titanium oxide, as the oxygen carrier with two solid fuels: a Mexican petroleum coke and a South African bituminous coal. These experiments involved testing at different fuel reactor temperatures, up to 1030°C, and different particle circulation rates between the air and fuel reactors. Previous results enabled modeling of the reactor system. In particular, it was possible to derive a correlation between measured operational data and actual circulation mass flow, as well as a model that describes the carbon capture efficiency as a function of the residence time and the char reactivity. Moreover, the kinetics of char conversion could be modeled and results showed good agreement with experimental values. The purpose of the present study was to complete these results by developing a model to predict the conversion of syngas with ilmenite in the fuel reactor. Here, kinetic data from investigations of ilmenite in TGA and batch fluidized bed reactors were used. Results were compared with the actual conversions during operation in this 10 kWth unit. Oil & Gas Science and Technology – Rev. IFP Energies nouvelles, Vol. 66 (2011), No. 2, pp. 181-191 Copyright © 2011, IFP Energies nouvelles DOI: 10.2516/ogst/2010023 Chemical Looping An Alternative Concept for Efficient and Clean Use of Fossil Resources La Boucle Chimique Un concept alternatif pour un usage propre et efficace des ressources fossiles IFP Energies nouvelles International Conference Rencontres Scientifiques d’IFP Energies nouvelles ogst100066_Berguerand 11/04/11 16:31 Page 181

Molar density of active solid phase (mol/m 3 ) τ Residence time in fuel reactor system (min) τ comp Time to reach complete conversion of an oxygen carrier particle (s)

INTRODUCTION
Chemical Looping Combustion (CLC) is a novel fuel conversion technology with inherent CO 2 capture.In the CLC process, the oxygen is directly transferred to the fuel via a circulating oxygen carrier in the form of a metal oxide, avoiding the presence of nitrogen in the exhaust flue gas, see Figure 1.
182 The Chemical Looping Combustion principle.
The oxidation of the oxygen carrier takes place in a socalled Air Reactor (AR) while it is reduced in the Fuel Reactor (FR).Denoting Me x O y and Me x O y-1 as the oxidized and reduced form of the oxygen carrier, the general reaction between the fuel and the metal oxide reads: (1) while the oxygen carrier oxidation in the air reactor is: In the fuel reactor, CO 2 and H 2 O are produced but these are never mixed with the nitrogen in the air as is the case during normal combustion with air.By condensing the steam, almost pure CO 2 can be obtained, which can later be transported to appropriate storage locations.At the outlet of the air reactor, the flue gas contains nitrogen and some unused oxygen.Thus, CO 2 is obtained avoiding active gas separation, as well as costs and energy penalties associated with gas separation.
Depending on the fuel and the oxygen carrier, reaction (1) is often endothermic while reaction (2) is exothermic.The total amount of heat resulting from reactions (1) and ( 2) is the same as for a normal combustion where the fuel is in direct contact with the air oxygen.
In CLC of solid fuels, the volatile compounds released from the fresh fuel can react according to reaction (1).However, the remaining char needs to be gasified in the fuel reactor using steam or CO 2 according to Equation ( 3): (3) The synthesis gases H 2 and CO can then react with the oxygen carrier according to: (4) Solid fuel CLC has been studied in laboratory units using different types of fuels and oxygen carriers [1][2][3][4][5][6][7][8][9][10][11][12][13][14].In particular, tests were conducted by Berguerand and Lyngfelt [15][16][17][18][19] and Markström et al. [20] in a 10 kW th unit at Chalmers University, totalizing almost 100 h of operation.The oxygen carrier used during these tests was ilmenite, an iron titanium oxide and the fuels a South African bituminous coal and a Mexican petroleum coke.More recently, two other pilots of 1 and 10 kW th were built and operated by Shen et al. [21] and Wu et al. [22] in China.
This work is based on results from batchwise introduction of fuel into the fuel reactor which gave the possibility to obtain complementary information compared to previous continuous tests.In recent work involving batch tests in this unit, direct evaluation of the oxygen demand associated with the syngas could be achieved.Moreover, the kinetics of char conversion in the fuel reactor was modeled as well as the loss of carbon to the air reactor as a function of the particle circulation.Here the main focus is made on modeling the conversion of syngas produced by char gasification.

EXPERIMENTAL BASIS 1.The 10 kW th Unit
A detailed description of the 10 kW th unit is provided in references [15,16].In this previous work, the fuel feed was continuous and here, the unit is generally operated in the / / same way except that the fuel is fed in batches.Figure 2 shows a general view over the whole reactor system: air and fuel reactors, particle filters, fuel feeding system, steam generator and water seal.
A schematic view of the different fuel reactor chambers and their dimensions is presented in Figures 3 and 4. Note that the arrows represent the directions of particle circulation.In particular, the fuel reactor is divided in three main chambers: -the low velocity section (LOVEL) subdivided into two chambers where the fuel is gasified and subsequently oxidized by the oxygen carrier; -the high velocity section (HIVEL) located below the riser through which both types of particles -but especially the unburnt char -are elutriated; the corresponding amounts depending on the operation/ fluidizing velocity.The reactor system has an internal recirculation loop via a smaller cyclone in order to increase the overall residence time of char particles.However, in the present work, the HIVEL chamber was fluidized at similar velocities as LOVEL; -the Carbon Stripper (CS) which has the purpose of separating char from oxygen carrier.The unreacted char is recirculated to the low velocity section while the oxygen carrier proceeds to the air reactor for regeneration.Moreover, two particle loop-seals separate the air and the fuel reactors, accommodating for the pressure differences between the reactors but also preventing cross-contamination of their effluents, see Figure 2. A third loop-seal, called "fuel reactor loop-seals" or FRPL, is located in the internal loop, between the leg of the small fuel reactor cyclone and the freeboard of the low velocity part LOVEL.

Oxygen Carrier and Solid Fuels
The oxygen carrier used in this work is an iron titanium oxide called ilmenite, which is extracted from a natural ore containing 40%wt ilmenite.It is provided by the Norwegian company Titania A/S and received in its reduced form FeTiO 3 .The iron/titanium molar ratio in this material is close to 1:1.The material used in the experiment has a pureness of 94.3% and a bulk density measured to 2100 kg/m 3 for the fresh ilmenite.However, studies indicate that the density of ilmenite decreases when the particles are subject to a CLC environment [23].
Oxygen carrier particles introduced into the reactor system were composed mainly of particles between 90 and 250 μm in diameter.The Particle Size Distribution (PSD) is shown in Table 1 below.
Note that ilmenite in this work refers to activated particles as these have previously been used during CLC operation in this unit and thus been subject to numerous consecutive redox cycles.Activation of ilmenite is detailed in [24].
In this work, results from two solid fuels are compared: a bituminous coal from the Republic of South Africa (RSA), and a petroleum coke from Mexico.The fuel analysis, for fuels as received, are shown in Table 2.  2 DATA EVALUATION AND PREVIOUS RESULTS

Data Evaluation
Criteria to assess the performance of operation in the 10 kW th unit need to be introduced.They include: -the oxygen demand, noted Ω OD , is an indicator of the degree of gas conversion in the fuel reactor and is defined as the fraction of oxygen lacking to achieve complete combustion of the gas produced in the fuel reactor.It depends on the contact and reactivity between the oxygen carrier and the reducing gases and is based on CH 4 , CO, H 2 and H 2 S; -the carbon capture efficiency η CC , which is defined as the ratio of gaseous carbon leaving the fuel reactor to the carbon containing gases leaving the system, i.e. air and fuel reactors.It reflects the residence time of particles in the fuel reactor and the separation efficiency in the carbon stripper; -the instantaneous rate of fuel conversion, based on the remaining amount of fuel in the bed at each time step of the reduction period.It is strongly temperature dependent.

Previous Results from Batch Operation
Results from batch operation in the 10 kW th unit are described in two earlier publications [19,20].In particular: -it can be shown in reference [19] that around 75% of the total oxygen missing for the conversion was associated with the fuel volatiles, confirming earlier estimation [18].Indeed, due to the fuel feed system with devolatilization occurring above the fuel reactor bed, the volatiles cannot be converted by the oxygen carrier particles.This means that the high oxygen demands usually obtained in testing with continuous feed in this unit are more a consequence of the reactor design than a problem inherent to the process [15][16][17][18]; -after batch introduction of fuel in FR and in presence of a particle circulation, the response in AR is displayed by a decrease in the oxygen concentration in the gas leaving AR.The shape of this oxygen consumption curve directly reflects the residence time distribution of the particles in FR.Then, using a model with N multistage well-mixed beds in series, it was possible to determine the residence time and residence time distribution of the particles in FR, for a number of operational cases with different particle circulation.Knowing the solids inventory in the fuel reactor, the actual circulation mass flow could also be determined [20].Moreover, the circulation mass flow could independently be correlated to measured operational data, i.e. pressure drop, temperature and gas flow in AR riser.This correlation could be used to determine the circulation mass flow in previous continuous testing [20]; Front view of the fuel reactor chambers (HIVEL is located right behind CS). 2) is the carbon stripper and 4) is the low velocity part LOVEL, divided into two chambers.The direction for the particle circulation are indicated.
Top view of the fuel reactor.2) is the carbon stripper CS, 3) is the high velocity part HIVEL and 4) is the low velocity part LOVEL, divided into two chambers.The directions for the particle circulation are indicated.
-using results for CO 2 capture efficiency in the 10 kW th unit and the parameters for the residence time distribution in FR above, a model was developed which could determine a mass-based reaction constant for char conversion.This mass-based reaction constant could also be determined independently from the conversion rates of char during the batch tests.There was good agreement between these two ways of assessing the rate constant, indicating that the model well describes the behavior of the unit.In other words, the model allows the prediction of the CO 2 capture from the particle circulation and char reactivity [20].
Here the idea is to complement these results by developing a model predicting the conversion of syngas in the fuel reactor for comparison with experimental results, using previous data on batch testing in the 10 kW th unit [19,20].

MODELING OF THE SYNGAS CONVERSION
Let us consider a bed of oxygen carrier particles with a total mass m tot and having an even distribution of char particles.When in contact with steam, the char particles will gasify to CO and H 2 .These syngases will further react with the oxygen carrier to form CO 2 and H 2 O.As the gas produced by gasification is assumed to be released proportionally to the mass, the gas flow F m at a level of the bed corresponding to a total mass m will be: (5) F 0 is the gas entering at the bottom, e.g.fluidizing gas, and κ is defined as: (6) where F g is the gas released from char gasification.Thus, κ = 0 would mean that there is no gasification occurring within the bed.Note that the gas flows are normalized to 0°C and 1 bar.
k F is introduced as a mass-based and volume-normalized rate constant indicating the reactivity, i.e. quantifying the ability of the ilmenite to convert the syngas at a certain fuel reactor temperature.Then, from a mass balance over a layer dm of the bed, the difference in partial pressure dp of the reactant, e.g.either of the two syngas species formed CO and H 2 , can be written as: (7) In Equation (7), the first term reflects the consumption of syngas reactant (CO or H 2 ) by reaction, the second the dilution by the release of syngas itself and the third the addition of reactant by syngas release, i.e. p g corresponds to the source of reactant through char gasification.Note that the derivation above is also valid for gaseous fuel, which corresponds to the case κ = 0 and p g = 0. Equation ( 7) assumes, in particular, a first order reaction for the consumption of reactant by the oxygen carrier and thus the following discussion is valid for the case of first order reactions.Integrating Equation (7) provides the general solution in Equation ( 8): (8) Here, p 0 is the partial pressure of reactant at the bottom of the bed and α a dimensionless number introduced as: In the 10 kW th unit, the bed is fluidized with steam, which means that p 0 = 0 and at the top of the bed, the partial pressure p in Equation ( 8) can be written: (10) Finally, the conversion of reactant at the top of the bed is derived from Equation (10) to: (11) During operation in the 10 kW th unit, the fluidizing flow of steam F 0 is constant at 3•10 -2 Nm 3 /min and the total bed mass in the fuel reactor is m tot = 5 kg.Moreover, κ can easily be calculated.Thus, to predict the conversion using Equation (11) and compare the result with experimental values during operation, the only missing parameter in the derivation above is k F .Two methods for obtaining k F are considered.
Method A utilizes experimental results from a lab unit operated with activated ilmenite and a bed mass m tot = 3 or 6 g together with a fluidizing gas flow of 9•10 -4 Nm 3 /min [25,26].In this study, syngas is added from below and there is no char gasification, i.e. κ = 0 and p g = 0. Thus, Equation (7) reduces to ( 12): (12) The solution to Equation ( 12) is: using the same notations as above.Finally the gas conversion reads: Moreover, the relationship between the degree of conversion X of an oxygen carrier defined as: (15) and the degree of mass-based conversion ω is: where R 0 is the actual oxygen transfer capacity of the oxygen carrier.Here, m ox resp.m red is the mass of a fully oxidized resp.reduced ilmenite particle.This means that from experimental plots giving γ against ω, the syngas conversion can be read.In [20], it was shown from testing with activated ilmenite in the 10 kW th unit that in order to satisfy both the heat and oxygen requirements in the fuel reactor, a maximum value for Δω of 2.1% is advisable.On the other hand, a lower value would correspond to an unnecessarily high particle circulation and thus a poorer CO 2 capture.Here, an interval of ω of 0.98-1 was used.Note that for values closest to 1, back-mixing considerably affects the results.To eliminate this effect, the cases where ω is above 0.995 are excluded from calculation with Method A. Then, extracting α from Equation ( 14) and introducing its value in (9), k F can be deduced using the values for m tot and F 0 for the lab unit.This k F is now used in Equation ( 9) to calculate α for the 10 kW th unit, and α is finally inserted in Equation (11) to obtain the conversion according to the model.
Method B utilizes the Shrinking Core Model (SCM) for the oxygen consumption within the ilmenite particles in combination with thermogravimetric analysis (TGA) results by Abad et al. (2010) [27].The SCM has been used extensively in order to model reaction kinetics of different type of oxygen carrier particles with CH 4 , CO and H 2 , normally with chemical reaction control only.The kinetic equation for the shrinking-core model for spherical grain is: (17) where τ comp is the time for complete conversion of the particle and calculated from: (18) Here, ρ m is the molar density of the active solid phase, r g the grain or particle radius, b the average stoichiometric coefficient for the reaction of solid with reacting gas and k the initial reaction rate constant.The unit for k is dependent upon the reaction order n, and is m/s for a first-order reaction [27].C is the concentration of reactant and can be considered as a constant due to the excess of reactant in relation to the oxygen carrier inventory during experiment.In [27], ρ m = 13.59 kmol/m 3 , r g = 125 μm and b = 1.45 for CO and H 2 .Moreover, R 0 was here 3.3% for activated ilmenite.
The reaction rate constant k is a function of temperature and hence fitted to an Arrhenius-type equation: From TGA experiment using syngas, the activation energy E and the pre-exponential factor k 0 can be obtained.Moreover, the reaction orders for H 2 and CO with activated ilmenite were determined to be 1 and 0.8 respectively [27].
An effective rate constant k eff has been introduced in [28] as: (20) where, p m is a mean partial pressure of the reactant in the bed.Thus, inserting Equation ( 16) in Equation ( 17) and differentiating with respect to "t", and combining with Equation (20) gives: (21) Moreover, from a mass balance over a bed layer dm and using Equations ( 12) and ( 20), k F and k eff are related through Equation ( 22): (22) where V n = 22.4 Nm 3 /kmol, p tot is the total pressure i.e. 1 bar, M O is the molar mass of oxygen and f OF is the stoichiometric ratio of oxygen consumed per mole of fuel oxidized, i.e. 1 for CO and H 2 .
Finally, combining Equations (17,20) and (21) gives: For a first order reaction, Equation ( 22) simplifies to: (24) Different values close to X = 0 can be chosen and the corresponding values of k F can then be used in Equation (9) to calculate ω, which is subsequently used in Equation ( 11) to obtain the conversions.

Calculation of kF
As seen in Section 3, the mass-based rate constant k F needs to be obtained first in order to determine the syngas conversions according to the model.
For hydrogen which exhibits a first order reaction with the ilmenite, methods A and B can be used directly.
Method A: Unpublished results from Leion et al. (2010) provide the ilmenite conversions of H 2 at different temperatures and using a bed of 3 g ilmenite with a fluidizing flow of 9•10 -4 Nm 3 /min [25], which means that m/F 0 = 0.2 ton•s/Nm 3 .Figure 5 shows the results.Then, using these conversions in Equation ( 14), α can be extracted.From the definition of α in Equation ( 9) and with m/F 0 above, k F is obtained in Nm 3 /(ton.s),see Table 3.
For comparison, results from operation in the same laboratory unit but using 6 g of the activated ilmenite and a fluidizing syngas flow of 9•10 -4 Nm 3 /min are also presented in Table 3.The values for hydrogen were taken from Figure 8 in [26].Results differ with a factor 3, as virtually the same conversions are obtained for twice as large a bed inventory.It should be said that evaluation using method A gives significant uncertainty when conversions are very high as is the case for hydrogen.Nevertheless, the data give the order of magnitude and are useful for comparison to method B.
Method B: From results on H 2 conversion in TGA experiment, the pre-exponential factor and the activation energy in Equation (19) were determined [27].Those were respectively k 0 = 6.2•10 -2 m/s and E = 65 kJ/mol.This enables calculation of the kinetics constant k at each operating temperature, using Equation (19).Finally, Equation (24) gives the corresponding k F .
There is a satisfying agreement in the results from methods A and B.
Note that a similar methodology is used in the calculation of k F at each temperature and these results are shown in Section 4.2 below.TABLE 3 kF (Nm 3 /(ton.s))for H 2 according to methods A and B at 950°C; A 1 using 3 g of ilmenite [25]; A 2 using 6 g of ilmenite [26]; B 3 using Equation ( 24); R 0 = 3.3% is used to convert X to ω For carbon monoxide, the order of reaction with activated ilmenite is 0.8, as obtained in [27].
For method A, a similar approach as above can be conducted, under the assumption of a first order reaction.Figure 6 shows the results from Leion et al. (2010) for the conversion against ω at different temperatures, using 3 g of ilmenite with a fluidizing flow of 9•10 -4 Nm 3 /min [25].
The results for k F are summarized in Table 4 for 950°C, and for both 3 and 6 g of ilmenite.In the case of 6 g of ilmenite, the values were taken from Figure 9 in [26].Here, results for 3 and 6 g seem to agree reasonably well.
In method B, the pre-exponential factor and the activation energy in Equation (19) were determined to 0.1 mol 0.8 m 0.4 s -1 and 80.7 kJ/mol respectively [27].Carbon monoxide conversion γ CO (gamma) against degree of mass-based conversion ω (omega) for different temperatures.From [25].
To calculate k F using Equation ( 23), a representative concentration C CO = p CO, m /RT (in mol/m 3 ) must be chosen.
Note that reaction (23) transfers the reaction constant k, which has the reaction order 0.8, to a constant k F assuming a first order reaction.Thus, k F is somewhat dependent on p CO, m or C CO and this should be considered when k F is being used.
For operation with petroleum coke in the 10 kW th unit, p CO, m /p tot = 0.02 is a fair approximation of the average carbon monoxide concentration in the bed of LOVEL [15][16][17][18].In the laboratory unit, for an inlet CO concentration of 50% and a conversion of 35% at 950°C, p CO, m /p tot = 0.4 can be chosen, see Figure 6 and [25].
Results from method B are summarized in Table 4, and k F is calculated using Equation ( 23) for the two concentration cases above.TABLE 4 k F (Nm 3 /(ton.s))for CO according to methods A and B at 950°C; A 1 using 3 g of ilmenite [25]; A 2 using 6 g of ilmenite [26]; B 3 using Equation ( 23); R 0 = 3.3% is used to convert X to ω p CO,m /p tot = 0.4 p CO,m /p tot = 0.02 As can be seen in Table 4, k F derived at p CO, m /p tot = 0.4 for method B agrees well with method A.

Syngas Conversion According to the Model
Here, conversion calculations are based on k F obtained for ω = 0.99.Since only results at 950°C are available in [26], the values of k F in method A will be calculated for the setup with 3 g ilmenite [25].For the carbon monoxide, the values of k F from method B will correspond to p CO, m /p tot = 0.02.
From a series of tests using petroleum coke presented in [19], the results for experimental conversions of H 2 and CO in the 10 kW th unit could be obtained.For corresponding operation, κ was calculated to 0.017.Moreover, the ilmenite inventory was 5 kg and the steam fluidizing flow 3•10 -2 Nm 3 /min, which means that m/F 0 = 10 ton•s/Nm 3 [19].
Thus, the predicted conversion γ H 2 could be determined from Equation (11), see Table 5.
Table 5 indicates that the two methods give fairly similar conversion predictions.High predicted conversions according to the model are due to the high values for k F and thus α, see Equation (11).This is consistent with the high hydrogen conversions obtained in the laboratory tests [25].However, a large deviation between measured and predicted conversions can be noted.
The results for CO are summarized in Table 6.For CO, the predicted conversions in Table 6 are somewhat higher for method B, which is because k F becomes higher at lower values of p CO, m .Again, the measured conversions are significantly lower than those predicted, although the difference is not as high as for hydrogen.

DISCUSSION
The conversion of the syngas from char gasification was modeled and results in Section 4.1 showed that there is a certain discrepancy with the experimental values obtained in the 10 kW th unit.Several factors that could explain this are discussed below.Firstly, an important point in the comparison of the different methods is that the bed is fixed in the TGA experiment and the mass transfer is different compared to the laboratory fluidized beds.Moreover, the reactivity of the ilmenite can vary depending upon the activation procedure.This can well explain the variations in the determination of k F seen in Tables 3 and 4. Nevertheless, the different approaches in determining k F give consistent results and, independent from these variations in k F , the model still predicts very much higher conversions than those actually measured.
Secondly, in the 10 kW th unit, there was an excess of fluidizing steam and the water-gas shift: CO + H 2 O → CO 2 + H 2 (25) was most likely pushed towards less CO and more H 2 , as confirmed by the higher conversions of CO compared to H 2 .As a consequence, the CO conversion agreed better with the higher model predictions.However, this shift from CO to H 2 does not explain the general difference in conversions between model and experiment.Finally, recent results by Linderholm and Cuadrat (2010) [29] obtained in the same unit suggest that the volatiles released in the early phase following fuel introduction are prone to form soot. Thus, a possible explanation to the deviations could be that soot was deposited above the bed and was gasified, producing H 2 and CO.A related explanation is if there are char particles not fully mixed into the bed but located downstream the bed, e.g in the loop-seal below the fuel reactor cyclone.

CONCLUSIONS
Results from batch tests performed in a 10 kW th CLC unit for solid fuels were used for comparison to model results.In these tests, ilmenite was the oxygen carrier and it was used in combination with two different fuels and different temperatures.
The idea with the present study was to obtain complementary information on the reactor system behavior.In previous work, it was possible to model the kinetics of the char conversion in the fuel reactor.Here, a model to predict the conversion of syngases CO and H 2 with ilmenite was derived.It involved either TGA or laboratory experiment for the calculation of rate constants for ilmenite.Results were compared with the actual conversions during operation in this 10 kW th unit.There is a discrepancy between predicted and measured gas conversions.A possible explanation could be soot formation from the fuel volatiles.

Figure 2
Figure 2 View of the 10 kW th unit: a) air reactor; b) air reactor riser; c) air reactor cyclone; d) fuel reactor.

TABLE 1
Particle size distribution of the ilmenite oxygen carrier (μm)

TABLE 6
Comparison between experimental and predicted CO conversions at different temperatures.a Experimental conversions in the 10 kW th unit.

N
Berguerand et al. / Chemical Looping Combustion of Solid Fuels in a 10 kW th Unit 189

TABLE 5
(11)arison between experimental and predicted H 2 conversions at different temperatures.aExperimental conversions measured in the 10 kW th unit.bFrom Equation(11)