Effect of the Moisture Content of Coal Blends on Coke Structure and its Apparent Gasification Kinetics with CO₂

Abstract

In this work, the influences of moisture content of coal on the structure and reactivity of cokes were investigated by blending different proportion of dry coal (with < 2 wt.% moisture) and wet coal (with ~10 wt.% moisture) and analyzing the gasification of the produced coke. The results indicated the coke formed from dry coal has the highest specific surface area and thinner pore walls. The results of isothermal thermogravimetric method show that the order of gasification reactivity of bulk coke from different proportion of wet coal is: 0 wt.% wet coal, 100 wt.% wet coal, 60 wt.% wet coal and 30 wt.% wet coal. In order to eliminate the influence of diffusion on the gasification reaction, coke with a particle size fraction of less than 48 µm was used for the non-isothermal gasification reaction. Results show that the gasification reaction curves of four samples are similar in the gasification process. It was concluded from kinetics analysis that the volume reaction model is well fitted with the experimental data. The activation energy with the volume reaction model is 191.9, 203.1, 190.1, and 190.8 kJ/mol. It was concluded that the moisture content of coal has little effect on the activation energy of the gasification, while the coke gasification kinetics is mainly determined by the coke pore structures which influence reaction surface.

1. Introduction

The blast furnace is still the major process for ironmaking from iron ore. Metallurgical coke is dominant material of the blast furnace which provides fuel for the blast furnace, reducing agent and permeability for both gas and liquid phases.^1,2) Coke gasification reaction is the most important factor in coke quality degradation, and these factors depend on the carbon structure, minerals, the morphology and connectivity of the pores.^2,3,4,5,6) In addition, the coking process of coal directly affects the performance of coke.

In recent years, the energy consumption and pollutant emission have been reduced by the technology of Coal Moisture Control (CMC, 5–6% moisture content) and Dry-cleaned and Agglomerated Precompaction System (DAPS < 2% moisture content).⁷⁾ Previous studies have demonstrated that reducing moisture content of coal can reduce heat consumption due to water evaporation.⁸⁾ Nomura and Arima has suggested that moisture can break through the plastic layer and cause uneven carbonization of coke.⁹⁾ In addition, a decrease in coal moisture leads to an increase in coal bulk density and increase productivity.^10,11) The increase in coal bulk density results in a more compact coke microcrystalline structure, which increases coke strength.¹²⁾ The increase in internal expansion pressure in the coking process caused by the increase in the bulk density of coal can be eliminated by increasing the proportion of low rank coal.¹¹⁾ As the main fuel for blast furnaces, it is desirable to predict coke reactivity from these factors. However, a lack of understanding of the effect of the moisture content of coal blends on coke structure and its apparent gasification mechanism is far from complete.

At present, microtopography and mercury porosimetry was used to the quantitatively investigate porous structure. Furthermore, the focus is to characterize moisture content of coal blends on coke gasification in CO₂ using isothermal and non-isothermal thermogravimetric analysis (TGA). Three kinetic models including the random pore model (RPM), volume reaction model (VM) and the grain model (GM) were used to determine the kinetic parameters, which has been successfully used in modeling gasification reactions of various carbonaceous materials. The research results of this paper provide a new understanding of the effect of the moisture content of coal blends on coke structure and its apparent gasification mechanism, and it can also provide guidance to the operation of coking process.

2. Experiment and Computational Methods

2.1. Material Preparation

The blended coking coal was obtained from a coking plant and their composition and properties are shown in Table 1. The coal sample was grounded and sieved and the particle size fraction used in this study was below 1 mm. The coking coal was divided into 2 parts, the moisture of one part was adjusted to be 10 wt.%, while the other part was dried at 105°C for 24 h under a flowing argon atmosphere (5 L/min). The first part was called wet coal, while the second part was called dry coal (with < 2 wt.% moisture). The samples of wet coal and dry coal were blended at 0 wt.% wet coal, 30 wt.% wet coal, 60 wt.% wet coal and 100 wt.% wet coal, respectively, which was labeled as C1, C2, C3 and C4. The mixture was blended for 30 minutes to homogenize completely and then stored hermetically for the subsequent experiments.

Table 1. Basic properties of blended coking coal.

Sample	Proximate analysis, wt.%, air dry basis				Gieseler parameters				Thickness of adhesive layer, Y/mm
	Vd	Ad	FCd	Sd	Maximum Fluidity (ddpm)	initial softening temperature (°C)	Resolidification temperature (°C)	Maximum fluidity temperature (°C)
Coal	26.37	9.63	64.3	0.42	1442	388	476	431	20.8

600 g of different moisture content of coal were respectively loaded into a square iron crucible and the coal charging volume and coal loading mass of samples were determined to obtain the coal material bulk density as shown in Table 2. An increase in coal moisture leads to a decrease in coal bulk density. The crucible was placed in a temperature-controlled muffle furnace. It is then heated in a two-stage process under nitrogen. In stage 1, the coal was heated at 5°C/min to 300°C, then at a rate of 10°C/min to 950°C. In stage 2, it was heated to 1200°C and hold for 3 h for coking. After the completion coking process, the furnace is shut down and naturally cooled to room temperature in nitrogen atmosphere.

Table 2. Bulk density of coal with different moisture in the same loading conditions.

Coal	C1	C2	C3	C4
Bulk density (g/cm3)	0.89	0.78	0.72	0.70

The morphology of the coking samples was observed by using Scanning electron microscope (SEM) (FEI Quanta-450) under a voltage of 15 kV. The coke with the particle size of about 4 mm was analyzed by the mercury intrusion method to characterize the pore size distribution and porosity. The specific surface area determined by porosimetry. A DMAX-RB X-ray diffractometer was used to assess the intrinsic carbon structures of the powder coking sample.

2.2. Gasification Tests

The Gasification experiments were conducted in a high-temperature furnace to investigate the mechanism of the reaction between the samples and CO₂. In this study, there are two methods, the isothermal techniques and the nonisothermal studies. In the isothermal test, coke gasification was carried out in a high-temperature furnace with program control. 6 g of bulk coke sample was loaded into a MoO hanging basket with height of 50 mm and a diameter of 40 mm. The deterioration of coke in the cohesive zone has a severe impact on the permeability of the blast furnace. According to results of the cohesive zone in the previous study,¹³⁾ the temperatures for isothermal experiments were selected to be 1000°C, 1100°C and 1200°C, respectively. The standard flow rate for CO₂ (purity≥99.99%) gas was 5 L/min. The reaction in CO₂ gas was in general run for 2 h. After 2 h, the gas was switched to nitrogen and the sample cooled down. In addition, in order to eliminate the influence of diffusion on the gasification reaction, the samples of coke were grounded and sieved and the particle size fraction used in non-isothermal study was below 48 μm. The reaction between the samples and CO₂ was investigated using a thermogravimetric analyzer (HCT-3, Beijing Henven Scientific Instrument Factory) to track the weight change of a sample with time at atmospheric pressure. 5 mg of initial sample were loaded into an Al₂O₃ crucible with height of 2 mm and a diameter of 5 mm in test. In this study, the reaction temperature was increased at a fixed rate of 5 K/min up to 1200°C under CO₂ at a gas flow of 100 mL/min CO₂ (purity≥99.99%) to avoid heat transfer limitations and minimize mass transfer effects.

The weight change during the isothermal and non-isothermal experiment was logged, and the coke conversion (x) was calculated by using the following equation:   

$$x=\frac{m_i-m_t}{m_i-m_f}$$(1)

Where m_i represents the initial sample mass; m_t is the weight of the sample at time t; m_f represents the remaining mass of the sample after gasification has been completed.

2.3. Kinetic Models

Coke gasification is one of the most common gas–solid heterogeneous reaction, and the rate of conversion was shown as below:   

$$\frac{dx}{dt}=k(T)f(x)$$(2)

Where f(x) is the mechanism function of conversion and k(T) is the apparent reaction rate constant, which represents the dependence of reaction rate on temperature, can be parameterized by the Arrhenius equation:   

$$k(T)=Aexp(-E/RT)$$(3)

where, E is the activation energy, A is the pre-exponential factor, and R is gas constant.

In this research, three nth-order kinetic models including the volumetric model (VM), grain model (GM) and random pore model (RPM), were chosen to describe the gasification process. The VM model assumed that the reaction of gasification take place in all active site, which does not investigate the structural transition of the particles during the reaction.^13,14)

When the rate limiting step of the reaction is controlled by the chemical reaction at the interface, the reaction rate is expressed as:   

$$\frac{dx}{dt}=k_{VM}(1-x)$$(4)

The GM model assumed that at beginning the reactant gas reacts with the external surface and moves progressively inside by degrees.^15,16) As the reaction proceeds, the surface area decreases nonlinearly with increased in reaction degree, the mechanism function of conversion can be expressed as follows:   

$$\frac{dx}{dt}=3k_{GM}{(1-x)}^{2/3}$$(5)

The RPM model initially proposed by Bhatia and Perlmutter, which accounts for the variation of pore characteristics owing to the reaction proceeds and coalescence of neighboring pores.¹⁷⁾ The mechanism function of conversion can be expressed as follows:   

$$\frac{dx}{dt}=k_{RPM}(1-x)\sqrt{1-\psi ln(1-x)}$$(6)

The symbol ψ in the above equation is pore structural parameter, defined as   

$$\psi =\frac{4\pi L_0(1-{\epsilon }_0)}{S_0^2}$$(7)

where L₀, ε₀ and S₀ correspond to the initial pore length, the initial solid porosity and initial pore surface, respectively.

In order to evaluate these models using the experimental results, Eqs. (5), (6), and (7) were linearized as shown in Eqs. (8), (9), (10), respectively:   

$$-ln(1-x)=k_{\text{VM}}t$$(8)

  

$$3[1-{(1-x)}^{1/3}]=k_{\text{GM}}t\mathrm{\hspace{0.17em} }$$(9)

  

$$(2/\psi )\left[{\left(1-\psi ln(1-x)\right)}^{1/2}-1\right]=k_{\text{RPM}}t$$(10)

3. Results and Discussion

3.1. Effect of the Moisture Content of Coal on Pore Structure of Coke

The quality of coke and gasification reaction is closely related to its pore structure, which accounts for 90% surface area.¹⁸⁾ Given in Fig. 1 are microtopography of different samples. It can be seen that the pore structure of coke changes significantly as the moisture content of coal increases. In Fig. 1 most of the pores of coke C1 are evenly distributed uniformly, with smaller dimensions and thinner pore walls. In addition, although the pores seem to be unconnected, studies have shown most of the pores in the actual coke are observed to be joined together when viewed from a three-dimensional perspective.¹⁹⁾ In the cohesive zone of the blast furnace, neighboring pores tend to coalesce and create new larger pores, which can improve the porosity of the coke and promote gas diffusion causing severe deterioration of metallurgical coke. Compared with coke C1, the coke C2, C3 and C4 are distinguished by relatively enlarged pores. However, it is formed that thin-walled macropore in C4, which will increase the specific surface area of the coke and deteriorate the strength during the production process of blast furnace. Back scattered electron image shows that inorganic minerals distribute in the carbon matrix. Figure 1 illustrates a typical aluminosilicate phase and an iron-containing phase with irregularly sharp and inserted into the coke matrix or filled in coke. Studies have shown that the aluminosilicate phase in the blast furnace also undergoes crystal changes with the sharp edges causing stress on the coke matrix and resulting in the formation of cracks.¹⁾ Iron, alkali and alkaline earth metals and their oxides have been known to catalyze coke gasification inside blast furnace.^20,21,22)

Fig. 1.

SEM–EDS micrographs of different samples. (Online version in color.)

Mercury porosimetry was used to the quantitatively investigate porous structure and the experimental data are shown in Fig. 2. The results show that the pore size distributions of cokes C1, C2, C3 and C4 is concentrated at 0–50 μm, 25–150 μm, 25–200 μm and 25–200 μm, respectively. The C1 has the largest proportion of pores with a size of 0–25 μm and the highest specific surface area. These pores are easy to connect and expand during the gasification reaction, which can also explain its high reactivity. In addition, it can be seen that the proportion of pores larger than 150 μm in the coke matrix increases with the proportion of mixed wet coal increases, expressed as a larger positive pore volume value. Mercury porosimetry was used to the quantitatively investigate porosity and specific surface area of porous structures. The error of this method is 3%. From Figs. 2(b)–2(d), It can be found that the specific surface area of C1 is significantly larger than C2, C3 and C4 coke, but the porosity and average pore diameter is opposite. The specific surface area of coke C2 is 6 m²/g, which is slightly lower than that of other coke. The C4 coke which is completely formed from wet coal has large average pore diameter and specific surface area. The porosity values of the cokes C2, C3 and C4 are quite similar. Combined with morphology analysis and pore properties, it indicates that pore structure is affected by moisture content of coking coal.

Fig. 2.

Pore structural parameters of samples: (a) pore size distribution, (b) specific surface area, (c) average pore diameter, (d) porosity. (Online version in color.)

The reason why the moisture content of coking coal greatly influences pore structure is considered as follows. The surface moisture between coal particles is a key factor affecting the density of coal.¹¹⁾ The resistance caused by the surface tension between the coal particles is greatly reduced as the surface adsorption water and interparticle water is depleted.²³⁾ Therefore, the density of dry coal is significantly higher than that of wet coal.

The blended coking coal was obtained from a coking plant. Coking coal is classified into low and high dilatation coal. The soften temperature of high dilatation coal and low dilatation coal is different.¹¹⁾ During the coking process, the low dilatation coal begins to soften at a lower temperature and has solidified when the high dilatation coal is still in a molten state (Fig. 3). When the sample is completely made from dry coal, the volume of coal expansion is larger than the volume between the interparticle space and the soften coal particles fill inter-particle space, eventually forming a firm bond. Previous studies have shown that the internal gas pressure increases sharply with the increase of bulk density.^9,11) Therefore, the generated gas cannot flow out of the outer layer through the plastic layer, for the close contact between the softened coke particles and is captured in the plastic layer. When the sample was blended 30 wt.% wet coal, the bulk density increases moderately. The solidified coal acts as a path for releasing gas, which reduces the internal gas pressure in the plastic layer. After 100% wet coal is coked, the proportion of large-sized pores is high. The water vaporization causes the volume to become large at the initial stage of coking process. The water vapor escapes outward through the gap between the coal particles. The pressure generated by the water vapor lead to the gap between the coal particles and an increase in the pore size.

Fig. 3.

Processes of coke pore formation. (Online version in color.)

3.2. Effect of the Moisture Content of Coal on Graphitization of Coke

X-ray diffraction profiles of the samples are shown in Fig. 4. It can be seen that the main mineral in these samples is quartz (SiO₂) and sillimanite (Al₂SiO₅), observed along with the micromorphology. According to carbon model by Franklin, the carbon matrix of coke is composed of graphite microcrystals.²⁴⁾ The characteristic XRD peak at ~25° and ~42° is assigned to the 002 and 100 peaks of graphite, respectively. From the results as shown in Fig. 4, the characteristic peaks of graphite almost overlap each other, which indicates that the moisture content of coking coal has little effect on carbon structure.

Fig. 4.

X-ray diffraction profiles of the samples. (Online version in color.)

3.3. Isothermal Gasification Kinetics of Bulk Coke

The evolution of carbon conversion curves with the different moisture content of coal blends at different temperature are given in Fig. 5. It can be seen obviously that the sample of C1 is typically more reactive in CO₂ than other samples at the same temperature, while the sample C2 has the lowest reaction rate. This is in agreement with the pore structure of coke. The pores of the sample C1 are easily connected and expanded during the gasification reaction, which results in an increased rate of reaction with the specific surface area increased. It was found that the moisture content of coking coal affects coke reactivity. Besides, as the reaction temperature increases, the carbon conversion of C3 and C4 are close to C1. With increasing temperature, the samples are gradually activated, resulting in a significant increase in weight loss. The rate of gasification reaction is controlled from the chemical reaction control to the diffusion control.

Fig. 5.

Evolution of conversion rate curves of coke at different temperature: (a) 1000°C, (b) 1100°C, (c) 1200°C. (Online version in color.)

The VM, GM and RPM models were used to determine the mechanism function of gasification. After fitting by Eqs. (8), (9), (10), the linear correlation of VM, GM and RPM is shown in Fig. 6. The gasification reaction curves of four samples are fit well by three models at different temperature, which indicates that there is no difference in the choice of the mechanism function of gasification of the bulk coke.²⁵⁾ From the Arrhenius plots of ln(K) against 1/T for the different samples shown in Fig. 7, kinetic parameters were obtained in Table 3. It is observed that the three different kinetic models do not substantially differ in characterizing the bulk coke gasification behaviors and the activation energy for coke is 70–130 kJ/mol at high temperatures, respectively. This result is consistent with the published activation energy of metallurgical coke.²⁶⁾ The lower activation energy of C1 may be due to the higher specific surface area. The coke activation energy of C2, C3 and C4 is similar, which indicates that the content of the moisture is not dramatically influenced the activation energy. As the reaction rate increases with the increase of temperature, diffusion becomes the key factor controlling the gasification behaviors of bulk coke.

Fig. 6.

VM, GM and RPM linearized models for different samples: (a) C1, (b) C2, (c) C3 (d) C4. (Online version in color.)

Fig. 7.

Arrhenius plots for the VM, GM and RPM models of different samples: (a) C1, (b) C2, (c) C3 (d) C4. (Online version in color.)

Table 3. Kinetic parameters of VM, GM and RPM models by fitting kinetic data of isothermal gasification.

Samples	VM		GM		RPM
	E/(kJ/mol)	lnk0	E/(kJ/mol)	lnk0	E/(kJ/mol)	lnk0	ψ
C1	90.36	−1.66	70.27	−3.46	70.59	−3.39	0.42
C2	123.87	1.07	119.67	0.35	117.80	0.20	1.23
C3	123.04	0.82	118.54	0.40	115.57	0.18	0.53
C4	119.15	0.66	110.66	−0.17	113.20	0.08	0.67

3.4. Non-isothermal Gasification Kinetics of Coke

In order to eliminate the influence of diffusion on the gasification reaction, coke with a particle size fraction of less than 48 μm was used for the non-equivalent gasification reaction. The evolution of carbon conversion curves is shown in Fig. 8. It can be seen obviously that the gasification reaction curves of four samples are similar in the gasification process. This indicates that pore structure is the key factor controlling the gasification process. The RPM, VM and GM models and calculation methods mentioned in section 2.3 were used to determine the most probable mechanism function of gasification. The evolution of conversion rate (dx/dt) curves and those calculated with VM, GM and RPM kinetics models are shown in Fig. 9. Results showed that calculated profiles with the VM model agree very well with the experimental data with the highest R² values, while that with GM model diverge far from the experimental results. The kinetic parameters (E, A and ψ) were obtained and shown in Table 4. It is observed that the activation energy for coke calculated by VM model are 191.9, 203.1, 190.1, and 190.8 kJ/mol.

Fig. 8.

Evolution of coke conversion curves with different moisture contents of coals under non-isothermal conditions. (Online version in color.)

Fig. 9.

Evolution of experimental conversion rate curves and those calculated with VM, RPM and GM kinetics models. (Online version in color.)

Table 4. Kinetic parameters of RPM, VM, and GM models by fitting kinetic data of non-isothermal gasification.

Samples	VM			GM			RPM
	E (kJ/mol)	A (min−1)	R2	E (kJ/mol)	A (min−1)	R2	E (kJ/mol)	A (min−1)	ψ	R2
C1	191.9	3.31E6	0.984	140.6	6.44E3	0.953	167.8	2.94E5	0.42	0.987
C2	203.1	1.05E7	0.982	163.2	6.06E4	0.963	177.2	6.46E5	1.23	0.980
C3	190.1	2.43E6	0.993	140.4	5.69E6	0.957	168.9	2.80E5	0.53	0.987
C4	190.8	2.72E6	0.996	140.7	6.18E3	0.973	169.1	2.84E5	0.67	0.994

The activation energy of cokes with and without wet coal blends is similar. In addition, X-ray diffraction profiles show that the moisture content of coal has little effect on carbon structure. The pore structure of coke C1 are evenly distributed uniformly, with smaller dimensions and thinner pore walls. As the gasification reaction proceeds, neighboring pores tend to coalesce and create new larger pores, which can improve the porosity of the coke and promote gas diffusion causing severe deterioration of metallurgical coke. Similarly, it is formed that thin-walled macropore in C4, which will increase the specific surface area of the coke and deteriorate the strength during the gasification. Generally, the reduction of pore diameter leads to an increase in connected pores, because dispersion exists in the pore-wall thickness of actual cokes. Thus, a pore-wall needs a certain thickness and a pore diameter also needs a certain size for reduction of connected pores.¹⁹⁾ Pores of C2 are isolated with divided thick pore-wall which reduces the coalescence of neighboring pores in the gasification reaction. Therefore, the pore structure is the key factor controlling the gasification process. The gasification reaction for metallurgical coke can be explained by the O-transfer mechanism.^27,28,29)   

$$C{O}_2+C_f=C(O)+CO$$(11)

  

$$(CO)=CO$$(12)

The pores are easily connected and expanded during the gasification reaction, resulting in the increase of specific surface area. The oxygen in carbon dioxide is more captured and then migrates into the carbon matrix to form a carbon (O) complex and increase apparent carbon reactivity.

4. Conclusions

Characterization of the carbon structure and reactivity of the coke has been carried out in order to establish effects of the moisture content of coal on coke characteristics. The results indicated the dry coal to coke has the highest porosity and thinner pore walls. As the coking coal moisture increases, the surface area and porosity increase. The density of coal is key factor controlling the pore structure. Excessive bulk density causes the pulverized coal to expand too much and the gas is trapped by the plastic layer to form a large number of tiny pores. Excessive moisture of coal forms a large gap in the pulverized coal particles, for the pressure of water vapor generated during the coking process.

The experimental data obtained in the isothermal thermogravimetric runs show that the coke made of dry coal is more reactive with CO₂ than other samples at the same temperature, while the sample containing 30 wt.% wet coal has the lowest reaction rate. The gasification reaction curves are fit well by three models, which indicates that there is no difference in the choice of the mechanism function of gasification of the bulk coke. The diffusion becomes the key factor controlling the gasification behaviors of bulk coke.

The gasification characteristics and kinetics of samples under non-isothermal conditions were analyzed to the influence of diffusion on the gasification reaction and ascertain the kinetic behaviors of different moisture -containing coke gasification. The gasification reaction curves of four samples are similar in the gasification process. The kinetic analysis showed that the VM model is well fitted with the experimental data. The activation energy with the VM model is 191.9, 203.1, 190.1, and 190.8 kJ/mol, respectively. It has little effect on the activation energy of the gasification reaction. This was thought to be a result of an in the number of active sites for gasification by the increased specific surface area. The oxygen in the carbon dioxide is more captured and then migrates into the carbon matrix to form a carbon (O) complex and increase apparent carbon reactivity.

Acknowledgements

The authors acknowledge the financial support of the National Key Research and Development Program of China (2017YFB0304300&2017YFB0304303), the National Science Foundation of China (51774032, 51974019), the Chinese Fundamental Research Funds for the Central Universities (FRF-TP-17-086A1) the National Science Foundation for Young Scientists of China (51804025).

References

• 1)   S.  Gornostayev and  J.  Härkki: Fuel, 85 (2006), 1047.
• 2)   K.  Li,  R.  Khanna,  J.  Zhang,  Z.  Liu,  V.  Sahajwalla,  T.  Yang and  D.  Kong: Fuel, 133 (2014), 194.
• 3)   S. M.  Shin,  I. H.  Jeong and  S. M.  Jung: Ironmaking Steelmaking, 45 (2018), 739.
• 4)   X.  Xing,  G.  Zhang,  H.  Rogers,  P.  Zulli and  O.  Ostrovski: Metall. Mater. Trans. B, 45 (2014), 106.
• 5)   F.  Meng,  S.  Gupta,  D.  French,  P.  Koshy,  C.  Sorrell and  Y.  Shen: Powder Technol., 320 (2017), 249.
• 6)   M.  Grigore,  R.  Sakurovs,  D.  French and  V.  Sahajwalla: ISIJ Int., 46 (2006), 503.
• 7)   S.  Wakuri,  M.  Ohno,  K.  Hosokawa,  K.  Nakagawa,  Y.  Takanohashi,  T.  Ohnishi,  K.  Kushioka and  Y.  Konno: Ironmaking Proc., Iron and Steel Society, Warrendale, PA, (1986), 303.
• 8)   R.  Wu and  G.  Xu: Chem. Ind. Eng. Prog., S1 (2012), 149 (in Chinese).
• 9)   S.  Nomura and  T.  Arima: Fuel, 87 (2008), 3240.
• 10)   P.  Cui,  K. l.  Qu,  Q.  Ling,  L. y.  Cheng and  Y. p.  Cao: Coke Chem., 58 (2015), 162.
• 11)   S.  Nomura,  T.  Arima and  K.  Kato: Fuel, 83 (2004), 1771.
• 12)   K.  Kato and  S.  Nomura: 5th Int. Cong. on the Science and Technology of Ironmaking Proc., Chinese Society of Metals, Beijing, (2009), 358.
• 13)   D.  Lacey,  J.  Bowen and  K.  Basden: Ind. Eng. Chem. Fundam., 4 (1965), 275.
• 14)   S. J.  Yuh and  E. E.  Wolf: Fuel, 63 (1984), 1604.
• 15)   A.  Calvelo and  R.  Cunningham: J. Catal., 17 (1970), 1.
• 16)   J.  Szekely and  J.  Evans: Chem. Eng. Sci., 25 (1970), 1091.
• 17)   S. K.  Bhatia and  D.  Perlmutter: AIChE J., 26 (1980), 379.
• 18)   G.  Liu,  P.  Benyon,  K.  Benfell,  G.  Bryant,  A.  Tate,  R.  Boyd,  D.  Harris and  T.  Wall: Fuel, 79 (2000), 617.
• 19)   Y.  Kubota,  S.  Nomura,  T.  Arima and  K.  Kato: ISIJ Int., 51 (2011), 1800.
• 20)   R. S.  Xu,  H.  Zheng,  W.  Wang,  J.  Schenk and  Z. L.  Xue: Energy Fuels, 32 (2018), 12118.
• 21)   S.  Gupta,  Z.  Ye,  B.  Kim,  O.  Kerkkonen,  R.  Kanniala and  V.  Sahajwalla: Fuel Process. Technol., 117 (2014), 30.
• 22)   S.  Gornostayev,  O.  Kerkkonen and  J.  Harkki: ISIJ Int., 45 (2005), 1.
• 23)   Z.  Rao,  Y.  Zhao,  C.  Huang,  C.  Duan and  J.  He: Prog. Energy Combust. Sci., 46 (2015), 1.
• 24)   R. E.  Franklin: Proc. R. Soc. Lond. A, 209 (1951), 196.
• 25)   M. M.  Sun,  J.  Zhang,  K.  Li,  K.  Guo,  Z. M.  Wang and  C. H.  Jiang: Int. J. Miner. Metall. Mater., 26 (2019), 1247.
• 26)   C.  Wang,  J.  Zhang,  K.  Jiao,  Z.  Liu and  K.  Chou: Metall. Mater. Trans. B, 48 (2017), 2428.
• 27)   M. J.  Roberts,  R. C.  Everson,  G.  Domazetis,  H. W. J. P.  Neomagus,  J. M.  Jones,  C. G. C. E.  Van Sittert,  G. N.  Okolo,  D.  Van Niekerk and  J. P.  Mathews: Carbon, 93 (2015), 295.
• 28)   D.  Zhao,  H.  Liu,  C.  Sun,  L.  Xu and  Q.  Cao: Combust. Flame, 197 (2018), 471.
• 29)   O. A.  Aladejebi,  B. J.  Monaghan,  M. H.  Reid,  M.  in het Panhuis and  R. J.  Longbottom: Steel Res. Int., 88 (2017), 1700039.