Modelling of gas conversion with an analytical reactor model for biomass chemical looping combustion (bio-CLC) of solid fuels

Manganese ores are promising oxygen carriers for chemical looping combustion (CLC), due to their high reac- tivity with combustible gases. In this work, a manganese ore called EB (Elwaleed B, originating from Egypt) is studied for its reaction rate with CH 4 , CO and H 2 and the data are used in an analytically solved reactor model. The reactivity of fresh and three used EB samples from previous operation in a 10 kW th pilot was examined in a batch fluidized bed reactor with CH 4 and syngas (50%CO + 50%H 2 ). In comparison with other manganese ores, the EB ore has a lower rate of reaction with CH 4 , while showing a significantly higher reactivity with syngas. Nevertheless, this manganese ore always presents a better conversion of CH 4 and syngas than the benchmark ilmenite. Mass-based reaction rate constants were obtained using a pseudo first-order reaction mechanism: 1.1 ⋅ 10 -4 m 3 /(kg ⋅ s) for CH 4 , 6.6 ⋅ 10 -3 m 3 /(kg ⋅ s) for CO and 7.5 ⋅ 10 -3 m 3 /(kg ⋅ s) for H 2 . These rate constants were used in an analytical reactor model to further investigate results from previous operation in the 10 kW th unit. According to the analytical model, in the 10 kW th operation, 98% of the char in the biomass fuels was gasified before leaving the fuel reactor, while the char gasification products (CO and H 2 ) have a 90% contact efficiency with the bed material. On the contrary, the volatiles have a much lower contact efficiency with the oxygen carrier bed, i.e. 20%, leading to low conversion of volatiles released. Thus, the results emphasize the importance of improving the contact between volatiles and bed material in order to promote combustion performance in the CLC process.


Introduction
Thermal energy from combustion provides the major power and energy demand worldwide, while the associated vast amount of emitted CO 2 is the main contributor to global warming. Carbon Capture and Storage (CCS) technologies can play an important role in the progress towards a low-carbon future, since neutral and even negative emissions can be realized [1]. As a promising category of CCS technology, Chemical Looping Combustion (CLC) burning fuels without mixing them with air has the advantageous feature of inherent CO 2 capture [2][3][4]. A schematic description of CLC is presented in Fig. 1, showing an air reactor, a fuel reactor and oxygen carrier particles circulating between these reactors. The oxygen carrier is normally metal oxides taking the oxidized form as MeO x and the reduced form as MeO x-1 [5]. In the case of using solid fuels, the fuels are decomposed to char and volatiles upon entering the fuel reactor at a typical temperature of 1073-1273 K, via pyrolysis reaction (R1). The resultant char is then gasified to CO and H 2 by the fluidization steam through reaction (R2). Subsequently, the gasification products (CO and H 2 ) together with volatile combustibles (CH 4 , CO and H 2 ) react with the oxygen carrier MeO x via reactions (R3)-(R5), while the oxygen carrier is reduced to MeO x-1 . After being conveyed to the air reactor, the reduced oxygen carrier is oxidized back to MeO x by the oxygen in air through reaction (R6). As can be seen in reactions (R3)-(R5), gases from the fuel reactor are mainly CO 2 and H 2 O, where the former can be easily captured after simple steam condensation. When using biomass fuels in CLC, i.e. bio-CLC, the captured and stored CO 2 will result in a net-flow of carbon out of the atmosphere, thus achieving negative CO 2 emissions [3,6]. The heat released from the CLC process is identical to that of conventional combustion, and can be used for power or heat generation [7][8][9][10]. Therefore, CLC is an innovative combustion technology for power and thermal energy generation with low energy penalty and inherent CO 2 capture.
Solidfuel→Char + Volatiles(CH 4 , CO, H 2 ) (R1) Char + H 2 O→CO + H 2 + Ash (R2) The oxygen carrier is a key for CLC, as it transfers the oxygen from combustion air to fuel in the fuel reactor [3]. Numerous works in the past decades have been dedicated to finding suitable oxygen carrier materials for CLC with gaseous and solid fuels [2,3]. In the case of solid Nomenclatures a ratio between oxygen demand for CH 4 in volatiles and the total oxygen demand for volatiles A c inner cross-section area of the reactor tube for modelling, m 2 b ratio between oxygen demand for CO in volatiles and the total oxygen demand for volatiles bio-CLC biomass Chemical Looping Combustion c ratio between oxygen demand for H 2 in volatiles and the total oxygen demand for volatiles C i,g concentration of gas i (i = CO or H 2 ) in the gasification products in the reactor for modelling, mol/m 3  oxygen carrier in the oxidized form MeO x-1 oxygen carrier in the reduced form n molar flow rate, mol/s n CI exponential factor considering the effect of circulation index CI on the modelling for overall gas conversion n i,g molar flow of gas i (i = CO or H 2 ) from char gasification in the reactor for modelling, mol/s n i,r consumption rate of gas i (i = CH 4 , CO or H 2 ) via gasoxygen carrier reactions in the reactor for modelling, mol/s ṅ out total dry gas molar flow after the batch fluidized bed reactor, mol/s n T exponential factor considering the effect of temperature T on the modelling for overall gas conversion p partial pressure, bar p i partial pressure of gas i (i = CH 4 , CO or H 2 ), bar fuels, inexpensive natural minerals and industrial by-products are preferred considering the inevitable oxygen carrier loss with ash removal. The most extensively studied and demonstrated oxygen carriers are ilmenite [11][12][13][14], iron ore [15][16][17][18], manganese ore [19][20][21][22][23][24][25][26] and CaSO 4 -based materials [27,28]. Among them, manganese ores have shown better reaction performance than ilmenite [22,23,29,30] and iron ores [23,30], and avoids the problem with sulfur emission from CaSO 4 [27,31,32]. In addition, there is a potential for gaseous O 2 release from manganese ores, a mechanism known as Chemical Looping with Oxygen Uncoupling (CLOU) [33], which means the fuel can burn with O 2 gas. This improves fuel conversion, as there is no need for steam gasification of the fuel. In this context, the cumulative operational experience with manganese ores in CLC pilots around the world exceeds 800 h [3]. Generally, most of the manganese ores studied have a small gaseous oxygen release [30,34]. Further, some ores show a noticeable promotion of char gasification as a result of catalytic effects of alkali impurities in the materials [23,35,36]. In several tests in continuous units, the conversion of intermediate components (CH 4 , CO and H 2 ) was significantly enhanced as compared to ilmenite [20,22,[24][25][26][37][38][39], which means less oxygen will be required for the oxy-polishing step [10,40,41]. On the other hand, high attrition rates have been observed in several cases, which would mean a shorter oxygen-carrier lifetime [42][43][44]. Adding foreign ions/supports [45,46] and sintering at high temperatures [21] can help to reduce attrition. In this context, a manganese ore, Elwaleed B, originating from Egypt, denoted as EB, with good reactivity and reasonably low attrition was demonstrated in a 10 kW th unit burning biomass-based fuels [22]. Using this manganese ore, a decrease of 8-10% in oxygen demand was reached as compared to ilmenite, while the estimated lifetime was around 370-830 h, which would be sufficient for industrial systems [10].
Modelling of reactors is a powerful tool for the development of CLC technology. A number of mathematical, multiphase Computational Fluid Dynamics (CFD) and even analytical models were developed and successfully applied to continuous reactors with a thermal power of 0.5 kW th -1 MW th in recent years [47][48][49][50][51][52][53][54][55]. These models use kinetic equations to describe the chemical reactions and semi-empirical/theoretical fluid dynamics to depict the gas/solid distribution in the reactors. Usually, the computational time cost and accuracy are the major concerns for these models, considering large numbers of calculating cells might be required. In comparison to CFD models, macroscopic mathematical model can give results within minutes of computational time [47,51], while keeping relevant information and giving good accuracy. The macroscopic model is more straightforward and has been well applied for the design and optimization of 10-100 MW th CLC plants [56,57]. Still, considering the limited experimental data available from larger units, and the fact that many parameters are unknown in this relatively new technology, the development of simplified modelling approaches could be valuable in order to identify importance of underlying parameters. Analytical modelling may in this regard be a useful tool. Such models are based on the analytical solution of differential equations considering reaction kinetics, fluid dynamics and mass balances [52,58]. An analytical model was demonstrated for gas conversion predictions in a 10 kW th unit [58] and a 100 kW th unit [52] for solid fuels, where the gas conversion was predicted as a function of oxygen carrier reactivity, solids inventory and gas flow. Analytical modelling, having the advantage of a high degree of transparency, could be an important tool for understanding and predicting CLC process better.
This work uses an analytical reactor model for the study of EB manganese ore oxygen carrier studied in recent 10 kW th operation [22]. Reactivity and reaction rate with the fuel components (CH 4 , CO and H 2 ) required by the model were determined by batch-reactor tests under fluidization conditions. Fluid dynamics and mass balances were considered, which together with the rates of gas-solid reactions and char gasification can predict gas conversion. From the analytical model, the rate of char gasification, contact efficiency of bed material with gasification products and volatiles in the 10 kW th pilot were estimated.

Oxygen carriers
Samples of fresh manganese ore "EB" and three used ones from a previous 10 kW th operation [22], as well as ilmenite from another 10 kW th operation campaign [59], were investigated in the current work. The used EB samples were from the initial, middle and final stages of the 42 h of operation in the 10 kW th unit [22] and are denoted as Us1, Us2 and Us3 in this work. Among the 42 h of operation in the 10 kW th reactor, 21 h involved the use of three biomass-based fuels at a temperature of 820-985 • C under steam fluidization [22]. As in the previous work [22], the fresh EB is composed of particles with a mean diameter of d50 = 183 μm after calcination, wet sieving and drying processes. The freshly calcined EB has a composition of around 24.8% Mn, 25.3% Fe, 8.1% Si and other minerals, as seen in Table 1. The ilmenite was used as a benchmark to compare with the EB ore. The original fresh calcined ilmenite before any tests mainly contains around 32.8% Fe, 25.0% Ti and 2.8% Si, as shown in Table 1. Since the ilmenite has undergone 21 h of operation with solid fuels in the 10 kW th unit at a temperature of 879-992 • C [59], it is in an activated state and the chemical composition may be different from the fresh material, e.g. due to iron migration to surface [60][61][62][63].

Fuels
The main gas-phase fuel components in CLC processes are CH 4 , CO and H 2 . Thus, in this work, CH 4 and syngas containing 50% CO and 50% H 2 were employed for the reactivity tests in the batch reactor. The idea with these experiments is to retrieve reaction rate data which can be used in the modelling. These two fuels were initially stored in highpressure cylinders. Prior to use, they were regulated to close to atmospheric pressure before being sent to the downstream reactor. The solid fuel used in the batch reactor experiments of this work is a char made by devolatilizing wood pellets at 1073 K under 100% N 2 environment in a batch reactor [64]. Therefore, the wood-pellet char generated is designated as "WPC", and it has around 93.5% carbon, <0.5% hydrogen and 3.0% oxygen. In the current work, the WPC char was added to a bed of EB particles fluidized by 100% N 2 , to investigate the rate of gas-phase oxygen release.

Batch fluidized-bed reactor setup
The reduction and oxidation cycles in the CLC process were performed by exposing the oxygen carrier alternatingly to fuel and oxidizing environment in a batch fluidized-bed reactor [65,66], as displayed in Fig. 2. The quartz reactor tube has a length of 820 mm and an inner diameter of 22 mm, which is surrounded by an electrically heated furnace. Bed materials were loaded on a porous quartz plate located inside the tube at around 370 mm above the bottom. Two thermocouples were placed at around 20 mm above and 10 mm below the porous plate to measure the temperatures for the bed and inlet-gas stream, respectively. To provide information about the fluidization state, the pressure drop over the bed was monitored by a transducer connecting between the top and bottom of the reactor [67,68]. Defluidization can be detected from loss of pressure fluctuations. Fluidization and fuel gases were introduced from the reactor bottom. These reacted with the bed material and then left via the reactor top. After steam condensation in the downstream cooler, the outlet gas was led to a gas analyzer (NGA 2000, Rosemount TM ) to measure the concentrations of CH 4 , CO, CO 2 , H 2 and O 2 as well as the volumetric flow at a rate of one sampling point every two seconds. Among the measured gases, H 2 sometimes shows a delay or an inaccurate concentration as a result of the internal adjustment in the analyzer. This will affect some of the calculated parameters but can be corrected by proper data processing methods [69], as discussed in Section 4 below. The temperature, gas concentration, volumetric flow and pressure drop were regularly registered in a computer connected to a data logger.

Reaction conditions
In all the tests in the batch reactor, the temperature was set to 1223 K. The bed composition and oxygen carrier reduction environment for different types of test are summarized in Table 2. In the case of gaseous oxygen release, 15 g fresh EB, Us1, Us2 or Us3 were used as the bed material, while the fluidization agent was 100% N 2 at a flow rate of 900 ml/min. In addition to this, there are several cases conducted with N 2 as fluidization gas but using WPC char as the reduction agent. In these cases, around 0.1 g WPC accompanied with 300 ml/min sweep N 2 were injected from the reactor top (i.e. the "Valve for solid fuel feed and sweeping gas" in Fig. 2). The oxygen release was characterized by the gaseous O 2 concentration detected in N 2 as well as the rate of oxygen carrier conversion in WPC experiments derived from CO and CO 2 produced. In the tests with gaseous fuels, CH 4 with a flow of 345 ml/min or syngas with a flow of 450 ml/min was used as reduction agent, while different amounts of fresh EB, Us1, Us2, Us3 or ilmenite were used. For CH 4 , 15 g oxygen carrier was used in the bed, whereas a mixture of 2 g oxygen carrier and 13 g sand was used for syngas. Less oxygen carrier was used in the latter case to avoid full syngas conversion and thus to acquire data useful for evaluation [70]. For both CH 4 and syngas, the reduction time was set to 20 s to lower the risk of defluidization and/or agglomeration caused by deep reduction [67,68]. The oxidation step was performed with 5% O 2 following 180 s of purge with N 2 .

Solid fuel pyrolysis
To compare with the experimental results in the recent 10 kW th operation [22], the same fuels "BP" and "SWC" were used for the analytical modelling. BP is steam cured wood pellets, called "black pellets", and contains 18.7% fixed carbon and 74.2% volatiles, while SWC is a commercial Swedish wood char and has 73.9% fixed carbon and 16.7% volatiles, as shown in Table 3.
From the fuel analysis, it is possible to predict the contents of gas components after pyrolysis using an empirical model [71], as seen in Eq. (1). In this simple model, the constants in the left square matrix were derived from large amounts of experimental data, while the right column matrix were based on fuel composition. The constants in the empirical model were obtained from coal pyrolysis and adapted to use for biomass-based fuels [71]. After solving the matrix equation, the pyrolysis products as displayed in the middle column matrix can be obtained, which gives the weight percentages of coke, tar and gas constituents. To be used in CLC process, the C 2 H 6 and tar calculated from Eq. (1) were further reformed and gasified in the model to CO and H 2 which added to the model-derived CH 4 , CO and H 2 representing the   volatile combustibles [51]. The other gas components (CO 2 , H 2 O, NH 3 and H 2 S) were not considered in the present work, as they are not combustibles or only present in trace fractions. As seen in Table 3, BP has a much higher fraction of the combustible gases CH 4 , CO and H 2 than SWC, which is consistent with the higher volatiles content in the proximate analysis. Meanwhile, CO and CH 4 are the more dominant gases in the volatiles according to the pyrolysis model, see Table 3, which might be a result of C 2 H 6 reforming and tar gasification.

Model description
In previous works, an analytical model for the fluidized-bed reactor was developed and adapted for simulating the gas conversion in a 10 and a 100 kW th unit [52,58]. This model considers an even distribution of oxygen carrier and char in the bed, as described in Fig. 3, where the total oxygen carrier mass is m tot . Fluidization and reactive gas comprised of H 2 O and volatiles has a total volumetric flow of V 0 , which enters the bed from the bottom. The volatile components are converted by the bed material, while their conversion is assumed to preserve the total inlet flow V 0 , meaning the volume expansion during volatiles combustion is not considered due to the low fraction of CH 4 as seen in Table 3. Thus, the outgoing gas at the bed surface is composed by two streams, i.e. V g and V 0 . The V g represents the volumetric flow of gas from char gasification, which can be a mixture of CO, H 2 , CO 2 and/or H 2 O depending on their extents of conversion. Assuming a linear increase of V g along bed height [52,72], the total gas flow V m at coordinate m can be written as Eq. (2).
Inside the reactor, char gasification as well as gas-solid reactions between fuel components (CH 4 , CO and H 2 ) and oxygen carrier were considered⋅H 2 O serves as a gasification agent and converts the char to H 2 and CO. The amount of CO and H 2 generated from char gasification dn i,g in the differential element dm was calculated by Eq. (3).
where C i,g is the concentration of gas i (i = CO or H 2 ) in the mixture of gas products from gasification before any reaction with the oxygen carrier bed. The fuel components CH 4 , CO and H 2 from both volatiles and char gasification are mainly converted by reacting with the oxygen carrier. In this work, a gas-solid reaction with a first-order reaction regime was used to describe the conversion of these components. Accordingly, the consumption, dn i,r , of gas i (i = CH 4 , CO or H 2 ) via gas-solid reactions in the dm element is given by Eq. (4).
where k F,i is a mass-based rate constant, determined by tests in the batch reactor, as seen in Section 4, and C i,m represents the concentration of component i at coordinate m in the bed.

Partial pressure of CH 4 , CO and H 2
In order to calculate the gas conversion for CLC process, the unconverted gases (CH 4 , CO and H 2 ) leaving the reactor are important to determine. Hence, based on a mass balance in the dm element, the differential equation Eq. (5) can be formulated and used to calculate the concentration change, dC i,m , in the differential element.
where dn i,g and dn i,r are the generation and consumption of gas i (i = CH 4 , CO or H 2 ) as a result of char gasification and reaction with the oxygen carrier, given by Eqs. (3) and (4), respectively. Further processing of Eq. (5) was done to get the partial pressure p i for gas i (i = CH 4 , CO or H 2 ) by applying the ideal gas law (pV = nRT) and atmospheric pressure in the reactor. In this case, Eq. (5) is converted to the following differential equation.
where κ is the ratio of ξV g and V 0 , and the symbol ξ indicates the fraction of char gasified in the fuel reactor and p i,g is the partial pressure of gas component i (i = CO or H 2 ) from char gasification. The value of ξ will be determined by fitting with experimental results from the previous 10 kW th operation [22]. Differential equation Eq. (6) has an analytical solution, which gives the partial pressure p i (m) of gas i (i = CH 4 , CO or H 2 ) at coordinate m and takes a general form as Eq. (7) below.
where p i,0 is the partial pressure of component i in gas entering the reactor, and the symbol α i is a dimensionless number, as shown in Eq. (8a).
where φ s,j is the contact efficiency of bed material and char gasification products (j = c) or volatiles (j = v), with φ s,c = 0.9 based on the high conversion of SWC char, with the oxygen demand as low as 2.6%, in the previous 10 kW th experiments [22] and agreeing with the results from previous works [52,53] and φ s,v is determined by fitting with the 10 kW th experiments [22]. Considering a reactor tube with A c as the inner cross-section area, Δp as the pressure drop over the bed while CI and T are the circulation index [73] and operation temperature, the α i can be calculated by Eq. (8b) considering the effects of solids circulation index and operation temperature [52] using exponential factors n CI and n T , respectively.
where the exponents n CI and n T are determined by fitting with experimental results. Initial values CI 0 and T 0 are from experimental averages, and g is the acceleration of gravity. The final gas conversion is obtained from the partial gas pressures at the bed surface according to Eq. (7). To do this, the general solution above is divided into two cases: (a) only char as fuel in the bed and (b) only gases as fuel in the bed. Therefore, the following Eqs. (9) and (10) were used to describe the partial pressure of gas i for char (i = CO or H 2 ) and volatile combustibles (i = CH 4 , CO or H 2 ), respectively, at the bed surface.

Gas conversion
From the expressions of partial pressure of gas component leaving the bed surface in Eqs. (9) and (10), the conversion, γ c,i for species i from char gasification (i = CO or H 2 ) and γ v,i from volatiles (i = CH 4 , CO or H 2 ), can be calculated using Eqs. (11) and (12), respectively.
The above obtained conversions for components i (i = CH 4 , CO and H 2 ) are further combined to predict the gas conversion for char and volatiles. Considering that the same amount of oxygen is needed for full oxidation of CO and H 2 , the gas conversion for char can be calculated as Eq. (13).
Similarly, the gas conversion for the volatiles can be obtained from the conversion of CH 4 , CO and H 2 using the ratios a, b and c for CH 4 , CO and H 2 in the volatiles, as seen in Eq. (14).
where the a, b and c denote the fraction of total oxygen demand relating to CH 4 , CO and H 2 . Thus, a + b + c = 1 and a, b and c were derived from the volatile combustibles in Table 3. Stoichiometrically, four moles oxygen are needed for complete combustion of one mole CH 4 , while one is needed for CO and H 2 . The ratio of oxygen required for CH 4 , CO and H 2 over the corresponding total amount in volatiles is denoted as a, b and c, respectively. For example, in the case of BP fuel in Table 3 (14) need to be further weighted together to form the overall gas conversion γ φ for the solid fuel. As shown in Eq. (15), the overall gas conversion depends not only on the oxygen demand ratio ψ for complete combustion of char in the fuel, but also on the char gasification fraction ξ.
where ψ is 0.36 and 0.88 for BP and SWC, respectively, based on the fuel analysis in Table 3, and the value of ξ is determined from the 10 kW th experimental results. With the overall gas conversion, the oxygen demand Ω OD , showing the ratio of theoretical amount of oxygen required for complete combustion of unconverted fuel-reactor gases (CH 4 , CO and H 2 ) over the oxygen needed for complete fuel combustion, is obtained from Eq. (16).

Inputs and outputs
The main inputs to the analytical model are reaction rate constants, reactor geometry, total pressure drop over the bed as well as the base operation conditions as shown in Table 4 below. Since the reaction rate were obtained with the batch-reactor tests at 1223 K, the base fuelreactor temperature T 0 in the analytical model was set to 1223 K. For the other parameters, the base values are the averages from stable operations in two recent 10 kW th campaigns [22,59], which are shown in Table 4 for different oxygen carrier and fuel pairs.
The outputs from the reactor model are the gas conversion of component i (i = CH 4 , CO or H 2 ) calculated by Eqs. (11) and (12) for char and volatiles, the gas conversion for char and volatiles via Eqs. (13) and (14) as well as the overall gas conversion from Eq. (15). Among these outputs, the overall gas conversion γ φ is the main focus in the Table 4 Base values to be input for the analytical model.

Data evaluation
Using the volumetric flow measured by the gas analyzer, the total gas molar flow ṅ out from the batch reactor was obtained by applying a factor from calibration. Then, the mass-based conversion ω of the oxygen carrier bed as a function of reaction time t is calculated by Eqs. (17) and (18) for CH 4 and syngas [34], respectively. The oxygen carrier conversion during the CLOU tests with WPC char is also calculated by Eq. (18) but with x H2 = 0, considering that 93.5% of this char is composed by carbon, see Section 2.2 above.
where t 0 corresponds to the time at the beginning of the reaction, M O and m ox are the molar mass of atomic oxygen and the mass of oxygen carrier used in the reactor, respectively. x i represents the fraction of gas i (i = CO 2 , CO or H 2 ) measured by the gas analyzer. Gas yield indicates the extent of fuel conversion to CO 2 , which was calculated from the measured gas concentrations, similar to previous works [74,75]. The gas yield, from CH 4 in the methane and from CO in the syngas, was obtained with Eqs. (19) and (20), respectively.
Using these gas yields, the mass-based rate constants for CH 4 and CO can be subsequently calculated by Eqs. (21) and (22), respectively. These equations consider first-order reactions between the oxygen carrier and the fuel components at atmospheric pressure [76,77], as the oxygen uncoupling with gaseous O 2 release is not significant for the EB ore as discussed in Section 5.1. The first-order assumption is in consistence with literature data [51,52,58,77,86] obtained in CLC reacting environment with various oxygen carriers, where the reaction orders are one or close to one for CH 4 , CO and H 2 .
where V CH4 and V syn are the inlet volumetric inlet flows of CH 4 and syngas, respectively. The symbol ε represents the volumetric expansion ratio for full combustion of CH 4 with ε = 2.
Since the rate constant for H 2 is sensitive to the measured H 2 concentration and the later can be incorrectly adjusted by the analyzer, the k F,H2 cannot be calculated with a similar method in Eq. (22). In this case, a molar ratio, ϑ, between the total oxygen consumed by H 2 and CO in the syngas was used to calculate the rate constant for H 2 [69], see Eq. (23). The value of ϑ was obtained by knowing the areas below the H 2 concentration curve and the CO concentration curve, i.e. integration of these concentration curves over reaction time, as well as the amount of H 2 and CO entering the reactor. Typically, the ϑ changed slightly in the range of 1.1-1.2 as a function of cycles, and this was considered during the calculation of rate constant for H 2 .

Progress of O 2 concentration in N 2 environment
The oxygen release from EB samples was tested in 100% N 2 environment immediately after the oxidation process with 5% O 2 . Meanwhile, the results for using sand as inert bed [78] as well as a CuO/Al 2 O 3 material [33], typical for CLOU, were employed for comparison purposes, as shown in Fig. 4. In the case of sand, the concentration of O 2 decreased rapidly to zero within 25 s from t = 15 to 40 s as the fluidization gas was switched from 5% O 2 to 100% N 2 . Because the sand is inert for oxygen release, the corresponding O 2 curve can be considered as a blank reference. In comparison to the sand bed, manganese ore has slower decrease in O 2 concentration and longer non-zero O 2 concentration duration (>100 s). This shows the capability of gaseous O 2 release from the manganese ore.
In Fig. 4, the O 2 release from Fresh EB, Us1, Us2 and Us3 is compared. The three used EB samples show a slightly higher O 2 concentration and a longer duration than the fresh material, although all the concentrations from these samples are lower than 0.5% after around t = 50 s. It is not clear if there is an activation of the EB ore for gaseous oxygen release, however, the capability of O 2 release is confirmed for both the fresh and used materials, consistent with the 10 kW th results [22]. The oxygen release from EB ore is also compared with a CuO/ Al 2 O 3 oxygen carrier [33] under similar reaction conditions. Obviously, the concentration of O 2 from the manganese ore is much lower, i.e. < 0.5% versus 2-4% at the same temperature of 1223 K [33,79]. Therefore, the manganese ore can be classified as a CLOU material with low oxygen release.

Rate of oxygen release
Under inert environment, e.g. in N 2 fluidization, the O 2 concentration from oxygen release may be limited by the thermodynamic equilibrium maximum value [33,79,80]. The addition of solid fuel consumes instantaneously the generated O 2 , removing the equilibrium barrier and thus shows the oxygen release rate in the absence of gas phase oxygen [33,34,81]. This is also more applicable for a real system, where it is expected that O 2 concentration in the bulk is very low [33,82]. A typical cycle using WPC as solid fuel in N 2 atmosphere is shown in Fig. 5, where the first 12 min corresponds to the reduction step which is followed by Fig. 4. Concentration of O 2 as a function of time after switching from 5% O 2 to 100% N 2 at a reactor temperature of 1223 K: comparison of Fresh EB, Us1, Us2 and Us3. Sand data from previous work [78] and CuO/Al 2 O 3 results from another work [33] are also presented for comparison. oxidation with 5% O 2 . When the WPC was introduced, peaks of all the measured gas components were detected, of which CH 4 and H 2 might be from the little residual volatiles in the char. The CO 2 and CO rising at the same time are a result of both a minor devolatilization and oxidation by the gaseous oxygen released from the oxygen carrier, while the peak of O 2 might be caused by the effects of solid fuel injection and/or the exothermic fuel combustion [9]. After the initial 1 min, only CO and CO 2 were observed during the reduction of oxygen carrier, which shows a continuation of solid fuel combustion by the insufficient O 2 from oxygen uncoupling. In the beginning of the oxidation period, high concentrations of CO 2 and small amount of CO indicate a low conversion of WPC during the previous reduction stage.
The rate of O 2 release is also described in Fig. 5 by the mass-based conversion ω of the oxygen carrier as a function of reaction time. Higher degree of reduction, i.e. a lower ω value, means that more oxygen has been transferred to the gas phase for fuel oxidation, thus a more reactive oxygen carrier with respect to CLOU. However, the EB manganese ore only reaches ω = 0.999. The average rate of oxygen release from the manganese ore is at a level of 0.01-0.015%/min, which is only 1% of the reaction with CH 4 and < 1‰ of the reaction with syngas as seen in Section 5.2.2. Thus, the gaseous O 2 release from the EB ore is small, and the CLOU effect can be neglected, which motivates the use of first-order reactions for EB with CH 4 and syngas. This is significant for the calculation of reaction rate constants and the analytical modelling under the assumption of first-order reaction regime.

Reaction progress
Gas concentrations in a typical redox cycle for the EB ore with CH 4 or syngas are presented as a function of reaction time in Fig. 6. Although the reduction time was set to 20 s, the gas concentration profiles last up to around 40 s, which is a result of gas back-mixing in the reactor, cooler and pipes leading to the gas analyzer, and a normal phenomenon in nonideal reactors [83]. In the case of CH 4 in Fig. 6(a), the main gas leaving the reactor is unconverted CH 4 which is as high as 70%, whereas CO 2 and CO are below 15%, indicating a low reactivity of EB towards CH 4 . There was no H 2 detected during the reduction, which might be a result of the full conversion of this intermediate component, considering its high reactivity. It is noted that although a significant fraction of CH 4 was not converted, there was no carbon deposition during the reduction step, because no carbonaceous gas was detected in the following oxidation. For the syngas in Fig. 6(b), the H 2 concentration is believed to be incorrectly adjusted by the gas analyzer, thus the initial H 2 concentration is slightly higher than CO, which was considered during the data processing, as seen in Section 4. Despite this, it is obvious that a much better conversion of syngas was achieved in comparison to CH 4 , as the major gas leaving the reactor was CO 2 while the total peak concentration of CO and H 2 was only around 25%. Thus, much higher conversion was reached with syngas, despite a much lower amount of active material, 2 g instead of 15 g in the bed. Considering the equal concentrations of CO and H 2 in the inlet syngas (50% CO and 50% H 2 ), the lower residual H 2 indicates a higher reactivity of this component with EB than CO. The slower oxidation after syngas reduction as compared to CH 4 reduction is a result of higher degree of EB reduction in the former case, i.e. 0.955 for syngas versus 0.995 for CH 4 . Agglomeration and defluidization of the bed were not observed during any cycle with CH 4 or syngas.

Rate of oxygen carrier conversion
The rate of oxygen carrier conversion, dω/dt, as a function of massbased conversion during the reaction with CH 4 or syngas is shown in Fig. 7. Generally, the rate increases significantly at the initial conversions and then remains at a relatively high value in the middle intervals of conversion, and finally declines quickly to a certain value as the fuel feeding is stopped. These rapid changes in the beginning and the end of reaction are attributable to the back mixing in the reactor as well as in the lines leading to and from the reactor [83].
In Fig. 7(a), the rate for fresh EB in various cycles with CH 4 and syngas is presented. Here the reaction rate increases slightly with rising cycle number. Despite this, it is hard to say if the material is activated or not in cycles, as differences are small. Nevertheless, it is clear that the syngas has much higher reactivity than CH 4 , as a 7-10 times higher value for dω/dt was achieved with syngas, leading to a higher degree of reduction of the EB material with syngas, reaching a ω as low as 0.955 as compared to above 0.995 for CH 4 . The effect of operation in the 10 kW th CLC unit [22] was also assessed by comparing the reaction rate of Fresh EB, Us1, Us2 and Us3 samples with CH 4 and syngas, see Fig. 7(b). For CH 4 , the Fresh, Us2 and Us3 show similar reactivity, whereas Us1 has a somewhat higher reaction rate. Also, for syngas the difference is small, with Fresh and Us3 showing slightly higher reaction rate. Therefore, it can be concluded that the EB ore has a stable reactivity with CH 4 and syngas.

Comparison of EB with ilmenite
In Fig. 8, the gas yields from CH 4 in methane and CO in syngas are shown as a function of conversion for EB and ilmenite. The gas yield falls with the increasing oxygen carrier conversion for both CH 4 and CO. In the case of CH 4 , EB ore has a gas yield of 0.27-0.15 as the conversion decreases from 1 to 0.996, which is around 20-40% higher than ilmenite. Using syngas as the fuel, the EB ore has a gas yield of higher than 0.76 from CO with the conversion varying in the range of 1-0.96, which is around 60% higher than the ilmenite. The higher gas yield for EB both with CH 4 and syngas is consistent with the lower oxygen demand for EB in the previous 10 kW th operation [22]. Fig. 9(a) shows the rate constant for CH 4 , CO and H 2 as a function of the mass-based oxygen carrier conversion for EB and ilmenite. CH 4 has the lowest reactivity with a rate constant smaller than 0.2⋅10 -3 m 3 /(kg⋅s) at all solid conversion levels for both EB and ilmenite, while the former is more reactive. In the case of CO and H 2 , the rate constant for EB is much higher than that for CH 4 , while the rate constant for H 2 is always higher than for CO. Compared to EB ore, the ilmenite shows a lower reactivity of all the gas components (CH 4 , CO and H 2 ).

Apparent rate constant
Using the oxygen carrier conversion of 0.98 for syngas and 0.998 for CH 4 as an example, the rate constant for CH 4 , CO and H 2 is plotted as a function of cycle number in Fig. 9(b), where the variation in reactivity for EB and ilmenite can be evaluated and further compared. For the EB material, the rate constant decreases slightly in the first 3-4 cycles and then becomes relatively stable for CH 4 , CO and H 2 . This behavior suggests a reactivity decrease and stabilization of the manganese ore during the first cycles with CH 4 and syngas [30]. On the contrary, for ilmenite a clear increase in the rate constant for CH 4 , CO and H 2 is seen in the cycles with both CH 4 and syngas. This activation behavior for ilmenite is commonly known in CLC processes using gaseous [60,61] and solid fuels [62]. For both EB and ilmenite, a stable rate constant is seen in the final cycles. Thus, EB has a higher reactivity than ilmenite with CH 4 , CO and H 2 .  As observed in Fig. 9(a), the rate constant changes with the oxygen carrier conversion. For the analytical model, adequate rate constant data are needed. According to previous works [84,85], an oxygen carrier conversion variation of around Δω = 0.02 should be assured in the CLC system to maintain sufficient oxygen transfer and heat balance. Therefore, the oxygen carrier conversion of 0.98 is considered for selecting the rate constants for CO and H 2 , whereas a value corresponding to the conversion of 0.998 is used for CH 4 because of the limited reduction degree of the oxygen carrier. As seen in Fig. 9(b), the reactivity decreased gradually to relatively stable values after some cycles. The rate constants from the last stable cycles were selected for use in the analytical model. The selected rate constants for CH 4 , CO and H 2 with EB or ilmenite are from cycle 10 with methane and cycle 6 with syngas and listed in Table 5.

Fitting the analytical model with 10 kW th experiments
A value of φ s,c = 0.90 has been proven reasonable for the contact efficiency of gasification products with the oxygen carrier in previous work [52] and is consistent with high conversion of SWC char in the 10 kW th experiments with an oxygen demand as low as 2.6% [22]. Thus, this value is also used in the current work for the analytical model. The parameters, ξ for gasification fraction of char and φ s,v for the contact efficiency of volatiles and bed, are determined by fitting with experimental results. According to the 10 kW th operation [22], the losses of char from BP or SWC are all lower than 2%, thus a high ξ should be expected, while a low φ s,v can be foreseen as big proportions of volatiles were unconverted. Moreover, these three parameters are not affected by the reactivity of oxygen carrier or fuel components, thus for all the solid fuels in the same reactor they should be kept at the same values. In this case, the values of ξ and φ s,v were determined as 0.98 and 0.20 for both BP and SWC with EB or ilmenite to have a good fitting with the experimental data. Using these values, the oxygen demand for BP and SWC is predicted and compared with experimental results, as seen in Fig. 10(a) for EB and Fig. 10(b) for ilmenite. The modelling results show somewhat underestimation for "EB + BP", "Ilmenite + BP", "Ilmenite + SWC" and overestimation for "EB + SWC" experimental data. And the tendency of oxygen demand variation can be well predicted by the analytical model, which means the values of these parameters (ξ, φ s,c and φ s,v ) have a good sensitivity over different operation conditions. Values for n CI and n T were obtained by fitting with experiments. Using the values for ξ and φ s,v determined in the section above, a sensitivity analysis was made to evaluate the effects of n CI and n T . From this, the values of n CI and n T for different fuel and oxygen carrier pairs can be determined, as seen in Fig. 11 using the EB ore and BP fuel pair as an example. Obviously, a constant gas conversion is calculated if the effect of oxygen carrier circulation is not considered, i.e. n CI = 0, which is also the case for temperature, i.e. n T = 0. Increase of either n CI or n T corresponds to increase the effect of solids circulation or temperature. Fig. 11(a) shows that the value of n CI = 3 leads to a reasonably good fitting with the experimental results [22]. Keeping n CI as 3, the use of different n T is shown in Fig. 11(b), where n T = 18 has the best fitting with experiments.
Similarly, the n CI and n T values were also determined for EB with SWC char and ilmenite with BP or SWC, see Table 6. Note that the n T for ilmenite and SWC was not determined due to the lack of enough experimental data. In return, the values in Table 6 can reflect the performance of fuels, oxygen carrier and their interactions in the 10 kW th unit. The ξ indicates the char gasification efficiency, while φ s,c and φ s,v show the contact efficiency of the bed with gasification products and volatiles. Accordingly, the char gasification has an efficiency of ξ = 0.98 in the 10 kW th unit, which is higher than the 0.70 for a bituminous coal obtained in the 100 kW th unit used in a previous investigation [52]. This is expected since biomass fuels normally have better gasification reactivity than coal [86]. The contact efficiency of volatiles and bed in the 10 kW th unit has a slightly higher value of 0.20 than 0.15 found in the bigger 100 kW th unit [52], which is likely caused by the improvement from adding the volatiles distributor [10,87] in the 10 kW th system.

Table 5
Apparent rate constants (m 3 /(kg⋅s)) for CH 4 , CO and H 2 in the analytical model.

Conclusions
In this work, reactivity data of a promising manganese ore oxygen carrier (EB) was determined in a batch fluidized-bed reactor and further used for analytical modelling as well as comparison from a 10 kW th pilot CLC reactor. It is confirmed that the EB ore has a low intensity of gasphase oxygen release, with<0.5% O 2 detected in N 2 atmosphere at 1223 K, which motivates the use of a first order reaction rate constant. Reactions with CH 4 and syngas indicate that the EB ore has a better reactivity than ilmenite. The stable rate constants for EB are 1.1⋅10 -4 m 3 /(kg⋅s) for CH 4 at the oxygen carrier conversion of 0.998, 6.6⋅10 -3 m 3 /(kg⋅s) for CO and 7.5⋅10 -3 m 3 /(kg⋅s) for H 2 at the oxygen carrier conversion of 0.98. Using these rate constants in the analytical reactor model, the overall gas conversion and oxygen demand were simulated, which fit well with the previous 10 kW th experiments. Accordingly, the biomass fuels had a high gasification fraction of 0.98 in the fuel reactor of the 10 kW th unit, while the contact efficiency of volatiles with bed material had a slightly higher value of 0.20 than previous results with gaseous and solid fuels in pilot operation.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Fig. 10. Comparison of oxygen demand for modelling results (filled symbols) and experimental results (void symbols) from previous 10 kW th operation [22,59]: (a) "EB + BP" and "EB + SWC" pairs, (b) "Ilmenite + BP" and "Ilmenite + SWC" (only data at 1260 K) pairs. Fig. 11. Determination of (a) exponent n CI for the effect of solids circulation and (b) exponent n T for the effect of fuel reactor temperature by fitting the overall gas conversion with the experimental data from the previous 10 kW th operation [22].

Table 6
Values of various parameters determined and used in the analytical model.