Integrating a Top-Gas Recycling and CO2 Electrolysis Process for H2-Rich Gas Injection and Reduce CO2 Emissions from an Ironmaking Blast Furnace

Introducing CO2 electrochemical conversion technology to the iron-making blast furnace not only reduces CO2 emissions, but also produces H2 as a byproduct that can be used as an auxiliary reductant to further decrease carbon consumption and emissions. With adequate H2 supply to the blast furnace, the injection of H2 is limited because of the disadvantageous thermodynamic characteristics of the H2 reduction reaction in the blast furnace. This paper presents thermodynamic analysis of H2 behaviour at different stages with the thermal requirement consideration of an iron-making blast furnace. The effect of injecting CO2 lean top gas and CO2 conversion products H2–CO gas through the raceway and/or shaft tuyeres are investigated under different operating conditions. H2 utilisation efficiency and corresponding injection volume are studied by considering different reduction stages. The relationship between H2 injection and coke rate is established. Injecting 7.9–10.9 m3/tHM of H2 saved 1 kg/tHM coke rate, depending on injection position. Compared with the traditional blast furnace, injecting 80 m3/tHM of H2 with a medium oxygen enrichment rate (9%) and integrating CO2 capture and conversion reduces CO2 emissions from 534 to 278 m3/tHM. However, increasing the hydrogen injection amount causes this iron-making process to consume more energy than a traditional blast furnace does.


Introduction
Traditional blast furnace (BF) iron making relies on carbon and contributes to over 70% of CO 2 emissions in the iron and steel industry [1]. In a blast furnace, coke is converted into a high-temperature CO gas and performs an exothermic reaction with iron ores, resulting in a large amount of CO and CO 2 leaving the furnace with top gas. Typically, every tonne of hot metal (tHM) produced from a traditional BF requires about 500 kg/tHM carbon and generates around 1.2 tonnes of CO 2 emissions [2,3]. Hence, there were various attempts for a clean iron-making process to reduce CO 2 emissions [4][5][6][7]. One of the approaches is using alternative reductants produced from renewable energies to replace carbon. BF operation with hydrogen as an auxiliary reducing agent was extensively investigated because of its specific advantages over CO [8][9][10][11][12]. Compared to CO reduction that generates CO 2 , reducing iron ores by hydrogen only forms water vapor. Kinetically, hydrogen enables a higher gas flow rate, and a faster reduction in iron ores and productivity than only CO does [13,14]. The higher thermal conductivity of hydrogen helps in heat transfer efficiency between solid and gas phases [15]. In addition, the hydrogen reduction in iron

Materials and Methods
Here, we studied the effect of H 2 -rich gas injection through BF tuyeres at the raceway position and/or shaft tuyeres, as shown in Figure 1. The BF is studied as the main subsystem when changing the hydrogen injection condition. The CCU and gas heating subsystem is used as a black box to provide necessary input information to BF. The BFG composition and flow rate from BF are the input for the CCU unit, and the CCU unit provides H 2 -rich reducing gas as the input to the BF. The operating conditions for this BF to produce 1 tonne hot metal (1 tHM) are listed in Table 1 and were kept constant throughout simulations. The overall CO content in the gas injectant was maintained at 200 m 3 /tHM. The productivity of this BF is 238 tHM per hour.
subsystem when changing the hydrogen injection condition. The CCU and gas heating subsystem is used as a black box to provide necessary input information to BF. The BFG composition and flow rate from BF are the input for the CCU unit, and the CCU unit provides H2-rich reducing gas as the input to the BF. The operating conditions for this BF to produce 1 tonne hot metal (1 tHM) are listed in Table 1 and were kept constant throughout simulations. The overall CO content in the gas injectant was maintained at 200 m 3 /tHM. The productivity of this BF is 238 tHM per hour.  As shown in Figure 1, hot oxygen-enriched blast and pulverised coal is injected through the tuyeres. The upper limit of oxygen enrichment rate for the blast was set at 14% to maintain the stable operation of the large-scale BFs. After drying and dust removal, some BFG is combusted in the hot blast stoves to heat cold blast, and in the gas heating device to provide high-temperature gas injectants. The rest of BFG enters the amine absorption CO2 capture unit to provide a CO2-rich stream that is processed in an electrochemical CO2 conversion unit to produce a CO and H2 stream containing, for example, 30% vol. H2 and 70% vol. CO. The CO2-lean stream from the top of the CO2 capture unit contained a mix of CO, H2, and N2 that is exported to other processes in the in-

Operating Parameters
PCI rate (kg/tHM) 137 Blast temperature, • C 1052 Humidity of hot blast, g/m 3 12.93 Top gas temperature, • C 161 Hearth injection temperature, • C 1250 Shaft injection temperature, • C 900 As shown in Figure 1, hot oxygen-enriched blast and pulverised coal is injected through the tuyeres. The upper limit of oxygen enrichment rate for the blast was set at 14% to maintain the stable operation of the large-scale BFs. After drying and dust removal, some BFG is combusted in the hot blast stoves to heat cold blast, and in the gas heating device to provide high-temperature gas injectants. The rest of BFG enters the amine absorption CO 2 capture unit to provide a CO 2 -rich stream that is processed in an electrochemical CO 2 conversion unit to produce a CO and H 2 stream containing, for example, 30% vol. H 2 and 70% vol. CO. The CO 2 -lean stream from the top of the CO 2 capture unit contained a mix of CO, H 2 , and N 2 that is exported to other processes in the integrated steel mill. We did not in this study consider additional separation of CO and H 2 from the N 2 in this stream for recycling back to the BF. Oxygen enrichment is required with H 2 -rich gas injection to provide heat to the BF and enrich BFG for CO 2 capture [35]. The BF can take another advantage from the CCU unit, as the electrolyser produces pure oxygen in another effluent stream. Besides BFG, a small amount of the basic oxygen furnace gas (i.e., BOFG) from steel making is usually combusted as fuel to a hot blast stove. Following assumptions of the BF, CCU unit, gas heating device and hot stove are made in this study: • Degree of indirect reduction R i depends on the reducing gas concentration of BF bosh gas, which is estimated by an empirical equation, as shown in Equation (1) [36]: where %(Reducing gas) is the proportion of H 2 and CO in the total amount of gas entering the bosh and shaft.
• For the amine absorption CO 2 capture unit: 30% monoethanolamine (MEA) concentration was used for CO 2 capture in this study. The capture unit recovered 90% CO 2 in BFG, and the CO 2 purity was >99%; the general thermal energy requirement for capture was assumed to be around 1000 kWh/tCO 2 (3.6 GJ/ tCO 2 ) [37][38][39]. Any additional CO 2 captured and not converted was assumed to be released as per current operation or could be sent to CO 2 storage routes, shown as CO 2 letdown in Figure 1.

•
The electrochemical CO 2 conversion unit was treated in the model as a simplified input-output model. Assumptions for the material and energy balance in the CO 2 conversion unit were based on laboratory demonstration data with additional inputs from literature sources. Briefly, the model was based on multiple two-cell vapour fed electrolyser stacks with the capacity to treat 50 tCO 2 per day; further details can be found in our other report [33]. The current density of the electrolyser was altered from 2.68 V at 0 A/m 2 to 3.59 V at 1862 A/m 2 to produce the H 2 -rich gas with different H 2 /CO compositions.

•
The electricity consumption for CO 2 conversion is proportional to the H 2 generation, which can be estimated as in Equation (2) [33]: where E conv is the power required for the CO 2 conversion unit, kWh/tHM; V H 2 gen is the amount of H 2 generated by the conversion unit, m 3 /tHM; • efficiency of the gas heating device was 85%; • The hot blast stove system uses two stoves on-gas and one stove on-blast, and the efficiency of the hot blast stoves was 75%.
As indicated in Figure 1, CO 2 capture and conversion units use renewable energy to avoid their own CO 2 emissions. The type of renewable power used by the industry depends on availability and cost, such as solar power [40,41]. Besides solar power, industries can use thermal-electrical materials to recover a large amount of waste heat in an integrated steel mill to provide electricity for CCU units [42]. In addition, using the lower heating value energy to generate electricity in the steel mill and the on-site power plant can help to minimise the renewable power periodic availability problem.
To achieve the objectives of this study, two mathematical models were developed. As a reducing agent, H 2 -rich gas needs to fulfil the thermodynamic requirement of the reduction reaction to capture oxygen in iron ores. H 2 -rich gas also needs to provide enough heat in the shaft for keeping an effective reduction process. First, hydrogen behaviour in different parts of the BF was analysed. A thermodynamic model for hydrogen reduction was built to determine hydrogen utilisation efficiency. This model provides a guideline of the proper hydrogen injection concentration. Then, a thermal balance model was used to limit the hydrogen injection temperature and volume.
The optimal hydrogen injection amount and position were determined by increasing the reduction potential in the coke consumption and CO 2 emissions, increasing gas utilisation efficiency and lowering the energy consumption. A static mass balance model of the BF was used to calculate the above parameters.

Thermodynamic Calculations of H 2 -Rich Gas Injection BF
The reduction behaviours of injected gas are discussed in different parts of the BF to determine the reducing gas utilisation.

Raceway
In the BF raceway, the main reactions considered in this study were carbon combustion, coke solution loss reaction between coke and CO 2 , water-gas reaction between coke and moisture in the hot blast, which can be described as shown in Equations (3)-(5): Solution loss reaction C (s) + CO 2 (g) → 2CO (g) ∆H 0 = 172,430 kJ/kmol (3) Water-gas reaction C (s) + H 2 O(g) → CO (g) + H 2 (g) ∆H 0 = 124,190 kJ/kmol (4) With excess coke existing in the BF bottom, it could be assumed that CO and H 2 combustion was negligible. This could be justified by the assumption that, if a small amount of H 2 reacts with O 2 to form water in the raceway, the generated water vapour reacts with coke and turn back to H 2 . Therefore, this process can be simplified as heating the hearth gas injectant, as shown in Equation (6): where T hearth is hearth gas injection temperature, • C; C p_hearth denotes specific heat capacity of the gas injected to BF, kJ/m 3 · • C; V hearth is hearth injection volume, m 3 /tHM; and ∆Q hearth is the sensible heat carried by the gas injected to BF, kJ/tHM. The primary reaction in the dripping zone is a direct reaction between coke and FeO. Hot gas containing H 2 and CO that passes through the cohesive zone reacts with molten FeO or semi-molten FeO to form H 2 O and CO 2 . As the temperature was over 1350 • C in the dripping zone, the amount of CO 2 was negligible due to the solution loss reaction. At the high-temperature zone over 1000 • C, some reduced FeO and Fe 3 O 4 were still in the solid state, and H 2 could pass through their surface. Almost all the H 2 O produced by H 2 reduction rapidly participates in water-gas reaction at the presence of coke to form CO and H 2 over 1273 K (1000 • C). Therefore, the reduction reaction in this section was essentially the direct reduction in iron by coke. H 2 injected through the tuyeres at the raceway mainly catalyses direct reduction and heats the molten or semi-molten burden.

Shaft Zone Temperature between 800 and 1000 • C
According to the thermodynamics of iron oxide reduction and dynamics studies, H 2 has better reducing capability than that of CO above 800 • C [43,44]. At the same time, the extent of coke solution loss reaction and the water-gas reaction was less than that in the higher temperature zone. In this temperature zone, H 2 reacts with various iron oxides to generate H 2 O, and the formed H 2 O is not gasified into H 2 by carbon completely. Therefore, it is the primary zone to improve H 2 utilisation efficiency.
The H 2 -rich reducing gas utilisation rate and volume requirement vary in different ferric oxides reduction stages. The iron oxides reduction reactions are at a nonequilibrium state in the BF. When the temperature is above 570 • C, the reduction in ferric oxides by CO and H 2 in the BF occurs in the following sequences: 1/2 Fe 2 O 3 → 1/3 Fe 3 O 4 (Stage I) → FeO (Stage II) → Fe (Stage III) [45]. The gas produced by the reduction in the latter stage is the reducing gas for the previous stage. Heat is gradually transferred to solid materials during the gas ascending. At the same time, part of reducing gas reacts with iron oxides and converts into CO 2 and H 2 O, and finally forms top gas at around 150 to 250 • C when leaving the BF. There are 25% of the total oxygen elements removed during the reduction of Fe 3 O 4 into FeO, and the remaining 75% of oxygen elements were removed in reducing FeO to Fe. Therefore, the reduction process from FeO to Fe is the key step. The required reducing gas amount is n kmol for CO, and m kmol for H 2 to produce 1 kmol iron. The value of n and m is the excess coefficient. The reduction reactions and thermodynamic parameters in Stage III for CO and H 2 are expressed as in Equations (7)- (12), and Equations (13)- (17), respectively [46]: ∆H 0 3CO = −13,190 kJ/kmol ∆G 0 3CO = −22,800 − 24.26T, kJ/kmol (8) where K is the reaction equilibrium constant; ϕ denotes the fraction of gas component. The minimal CO required for Stage III is described in Equation (11): The utilisation efficiency of CO in Stage III, η 3CO , is described in Equation (12): ∆H 0 3H 2 = 28,010 kJ/kmol The minimal H 2 required for Stage III is calculated as in Equation (16): The utilisation efficiency of H 2 in Stage III, η 3H 2 is described in Equation (17): According to the theoretical thermochemical calculations, 50% of H 2 and CO participates in the water-gas shift reaction Equation (18) between 600 and 1400 • C [47]. Therefore, the heat consumed by the water-gas shift reaction at temperatures above 820 • C is balanced by the heat generated at the temperature below 820 • C, as calculated by Equation (19). However, H 2 promotes iron ore reduction by CO via the water gas shift reaction when the temperature is over 820 • C [48]. The CO 2 generated reacts with H 2 to reform CO, which participates in FeO reduction reaction again and improve the utilisation efficiency of CO.
Water-gas shift reaction The heat effect of FeO reduction by H 2 and CO gas is calculated by Equation (20): where X i is the proportion of CO or H 2 in the reducing gas entering the BF shaft. The gas produced by the reduction in Stage III is the reducing gas for Stage II. The reduction reactions and thermodynamic parameters in Stage II for CO and H 2 are expressed in Equations (21)-(28): The minimal CO and H 2 volume required for Stage II is calculated as shown in Equations (29) and (30), respectively: The utilisation efficiency of CO and H 2 in Stage II is calculated as shown in Equations (31) and (32), respectively: In the first stage of iron ores reduction, the transformation of Fe 2 O 3 to Fe 3 O 4 is very rapid due to the very high equilibrium constant of Fe 2 O 3 reduction above 600 K, as shown in Equations (33) and (35): The gas produced by the reduction in Stage II provides the reducing gas for Stage I. These reactions only require a low concentration of reducing gas to proceed. The minimal CO and H 2 volume required for Stage I is shown in Equations (37) and (38), respectively. With utilisation efficiency close to 100%, Fe 2 O 3 reduction is an irreversible reaction.
The utilisation efficiency of CO and H 2 in Stage I is described by Equations (39) and (40), respectively: The overall gas utilisation efficiency for H 2 -rich reducing gas in the BF is calculated as in Equation (41) below: Assuming the water generated in the Fe 2 O 3 reduction is reacted with CO, in which H 2 performs only as a catalyst of CO reduction of Fe 2 O 3 . The water in top gas is determined by H 2 utilisation efficiency in FeO and Fe 3 O 4 , which was calculated as in Equation (42): Since FeO reduction is the key step, the theoretical overall H 2 utilisation efficiency was calculated as shown in Equation (43). The highest theoretical H 2 utilisation efficiency can be obtained with the minimal H 2 requirement value on the basis of the thermodynamic requirement in Stage III, and this highest value is determined by temperature. Due to thermal restrictions and excess H 2 injected, the actual gas utilisation efficiency can only approach this theoretical value. The actual thermodynamic utilisation efficiency of H 2 is a function of the amount of H 2 introduced to the BF, as shown in Equation (43): As the FeO reduction is the key step in the indirect reduction process, the thermodynamic requirement of gas entering the BF shaft to produce 1 tHM is calculated as in Equations (44) and (45): where V bosh_shaft is the amount of gas raised from BF bosh after direct reduction and the gas injected through the shaft tuyeres, m 3 /tHM; [Fe] HM is the proportion of iron content in hot metal; R d is the degree of direct reduction.

Thermal Calculations of H 2 -Rich Gas Injection BF
As the heat carrier, the H 2 -rich gas injected through raceway tuyeres needs to compensate for the required energy in the lower furnace and maintain the theoretical combustion temperature at a reasonable range. The gas injected through the shaft also needs to satisfy the heat requirement in the upper furnace. The energy of H 2 -rich gas includes the oxidation heat release from the iron ore reduction and sensible heat. The oxidation heat release depends on gas utilisation efficiency and gas composition. The injection temperature determines the sensible heat. Thermal calculations for determining the amount of H 2 -rich gas were developed by a static mass and energy model of the iron-making process.
The thermal balance for this iron-making process is developed in the lower and upper furnaces, divided by the shaft gas injection position. In this work, the lower furnace included BF raceway, dripping zone, and cohesive zone. The thermal balance of the lower furnace is shown in Equation (46) below: (46) where the heat income in the lower furnace includes: Q cc = combustion heat of coke and pulverised coal in front of tuyeres, kJ/tHM; Q blast = sensible heat of the hot blast, kJ/tHM; Q hearth = sensible heat of H 2 -rich gas injection to the hearth, kJ/tHM; Q coke = heat of the coke brings to the lower part of BF, kJ/tHM; Q ore = sensible heat of the iron ores into the lower part of the BF, kJ/tHM; and the heat expenditure includes: H CO2 = heat consumption of solution loss reaction due to the possible CO 2 in the hearth injection gas, kJ/tHM; H H 2 Od = heat consumption of water decomposition in front of tuyeres, kJ/tHM; H PCde = heat consumption of pulverised coal decomposition in front of tuyeres, kJ/tHM; Q bosh = heat brought to the shaft by bosh gas, kJ/tHM; H dA = heat consumption by direct reduction of alloy element, kJ/tHM; H dFe = heat consumption by direct reduction of FeO, kJ/tHM; H S = heat consumption by desulphurisation, kJ/tHM; Q HM = sensible heat of hot metal, kJ/tHM; Q slag = sensible heat of slag, kJ/tHM; and Q loss_l = heat loss in the lower furnace, kJ/tHM. With H 2 -rich gas injection to BF hearth, the raceway adiabatic flame temperature (RAFT) is calculated as in Equation (47): where Q coke is the heat brought to the raceway by coke, kJ/tHM; V raceway i is the gas volumes of H 2 , CO and N 2 in the raceway, m 3 /tHM, respectively.
The energy input and output of the BF shaft can be expressed as in Equations (48) and (49), respectively: where r ico and r iH 2 are the degree of indirect reduction by CO and H 2 , as shown in Equations (53) and (54), respectively:

Results of the Thermodynamic Model
At a medium oxygen enrichment rate (9%), the nitrogen content in the BF shaft is calculated at around 35%. Figure 2a shows the heat effect results based on Equation (20). The FeO reduction reaction transforms from an exothermic into an endothermic process when H 2 content increases to 25% around 900 • C. A similar phenomenon shows up when the H 2 reaches 20% around 1000 • C. With more H 2 participating in the reduction at high temperatures, it causes a severe negative effect on the thermal energy supply to the BF. Therefore, the shaft gas injection temperature should not be too high to reduce its endothermic heat effect and require less preheating in the gas preheating device. The higher oxygen enrichment (less N 2 content) enables higher H 2 content in the BF, as shown in Figure 2b. It is suggested that with 20% N 2 entering the BF shaft, the H 2 content should be lower than 25% to avoid too much heat consumption by its reduction reaction.
According to Equations (11), (16), (29) and (30), the theoretical minimal H 2 and CO requirement with temperature are shown in Figure 3. From this figure, Stage III for iron ores reduction requires more H 2 when the temperature is over 625 • C. Stage III requires more CO over 650 • C. In this study, since the gas injection temperature is kept above 820 • C to ensure the high reducing capability of H 2 , the thermodynamic key step for iron ores reduction is Stage III, and the other stages are proceeded with excess reducing gas. Compared with Fe 3 O 4 , FeO reduction requires more reducing gas to proceed, which determines the minimal amount of gaseous mixture. Fe 3 O 4 and Fe 2 O 3 reductions are carried out with excess reducing gas. In addition, the amount of H 2 required for FeO reduction at high temperatures is less than the amount of CO required in the reaction. Thermodynamically, injecting H 2 content to replace some amount of CO would reduce the total amount of gas mixture and reduce the fuel requirement for the gas preheating. However, CO content in the BF should be enough to meet its thermal condition.
According to Figure 4, gas utilisation efficiency for H 2 and CO was similar in Stage II when the temperature was above 820 • C, but H 2 utilisation efficiency was much higher than that for CO in Stage III. Although the utilisation efficiency of CO decreases as temperature increases, the FeO reduction reaction would still be promoted with increasing H 2 content due to the effect of the water-gas shift reaction.
On the basis of the calculation from Equation (41), gas utilisation efficiency at different H 2 contents in the reducing gas for FeO and Fe 3 O 4 reduction is shown in Figure 5.
BF. Therefore, the shaft gas injection temperature should not be too high to reduce its endothermic heat effect and require less preheating in the gas preheating device. The higher oxygen enrichment (less N2 content) enables higher H2 content in the BF, as shown in Figure 2b. It is suggested that with 20% N2 entering the BF shaft, the H2 content should be lower than 25% to avoid too much heat consumption by its reduction reaction. According to Equations (11), (16), (29) and (30), the theoretical minimal H2 and CO requirement with temperature are shown in Figure 3. From this figure, Stage III for iron ores reduction requires more H2 when the temperature is over 625 °C. Stage III requires more CO over 650 °C. In this study, since the gas injection temperature is kept above 820 °C to ensure the high reducing capability of H2, the thermodynamic key step for iron ores reduction is Stage III, and the other stages are proceeded with excess reducing gas. Compared with Fe3O4, FeO reduction requires more reducing gas to proceed, which determines the minimal amount of gaseous mixture. Fe3O4 and Fe2O3 reductions are carried out with excess reducing gas. In addition, the amount of H2 required for FeO reduction at high temperatures is less than the amount of CO required in the reaction. Thermodynamically, injecting H2 content to replace some amount of CO would reduce the total amount of gas mixture and reduce the fuel requirement for the gas preheating. However, CO content in the BF should be enough to meet its thermal condition.  According to Figure 4, gas utilisation efficiency for H2 and CO was similar in Stage II when the temperature was above 820 °C, but H2 utilisation efficiency was much higher than that for CO in Stage III. Although the utilisation efficiency of CO decreases as temperature increases, the FeO reduction reaction would still be promoted with increasing H2 content due to the effect of the water-gas shift reaction.  According to Figure 4, gas utilisation efficiency for H2 and CO was similar in Stage II when the temperature was above 820 °C, but H2 utilisation efficiency was much higher than that for CO in Stage III. Although the utilisation efficiency of CO decreases as temperature increases, the FeO reduction reaction would still be promoted with increasing H2 content due to the effect of the water-gas shift reaction. On the basis of the calculation from Equation (41), gas utilisation efficiency at different H2 contents in the reducing gas for FeO and Fe3O4 reduction is shown in Figure 5.  Figure 5a presents the gas utilisation efficiency of FeO reduction. Due to the endothermic reaction of the H2 reduction, utilisation efficiency of H2-rich reducing gas at 1000 °C for the FeO reduction increased from 23% with no H2 to around 30% with 50% H2 by the equal interval. When H2 content was less than 30% in reducing gas, gas utilisation efficiency in FeO reduction decreased with the increase in temperature. In contrast, gas utilisation efficiency rose with temperature when H2 content is more than 40% in reducing gas. Hence, H2 content in the reducing gas should not be too low to hinder improvement in the gas utilisation efficiency. As shown in Figure 5b, since Fe3O4 reductions by CO and H2 are endothermic, gas utilisation efficiency increases with temperature. The effect of increases in H2 content for Fe3O4 reduction is less than FeO reduction in terms of gas utilisation efficiency. At 1000 °C, gas utilisation efficiency increases by less than 4% when H2 content increases from 0% to 50%. The shaft gas injection temperature should be higher than 820 °C to promote gas utilisation in FeO and Fe3O4 reduction, especially focusing on FeO reduction.

BF Simulation Conditions and Validation
Without gas injection, the model developed in this work can be used for a traditional BF. The measured data collected from a 2500 m 3 BF were used to validate this proposed model. The comparison of the industrial data and the model predictions is summarised in Table 2. The coke rate and top gas components were compared because they are essential measurable parameters that indicate the overall performance of an iron-making BF. The chemical composition of raw material data is shown in Tables 3-5. In general, results in this simulation show a similar trend as in practice, and the model was capable of estimating the overall iron-making process. The slight difference in top gas composition is because industrial data contain a slight amount of O2 and CH4 in top gas, which does not exist in this model.  Figure 5a presents the gas utilisation efficiency of FeO reduction. Due to the endothermic reaction of the H 2 reduction, utilisation efficiency of H 2 -rich reducing gas at 1000 • C for the FeO reduction increased from 23% with no H 2 to around 30% with 50% H 2 by the equal interval. When H 2 content was less than 30% in reducing gas, gas utilisation efficiency in FeO reduction decreased with the increase in temperature. In contrast, gas utilisation efficiency rose with temperature when H 2 content is more than 40% in reducing gas. Hence, H 2 content in the reducing gas should not be too low to hinder improvement in the gas utilisation efficiency. As shown in Figure 5b, since Fe 3 O 4 reductions by CO and H 2 are endothermic, gas utilisation efficiency increases with temperature. The effect of increases in H 2 content for Fe 3 O 4 reduction is less than FeO reduction in terms of gas utilisation efficiency. At 1000 • C, gas utilisation efficiency increases by less than 4% when H 2 content increases from 0% to 50%. The shaft gas injection temperature should be higher than 820 • C to promote gas utilisation in FeO and Fe 3 O 4 reduction, especially focusing on FeO reduction.

BF Simulation Conditions and Validation
Without gas injection, the model developed in this work can be used for a traditional BF. The measured data collected from a 2500 m 3 BF were used to validate this proposed model. The comparison of the industrial data and the model predictions is summarised in Table 2. The coke rate and top gas components were compared because they are essential measurable parameters that indicate the overall performance of an iron-making BF. The chemical composition of raw material data is shown in Tables 3-5. In general, results in this simulation show a similar trend as in practice, and the model was capable of estimating the overall iron-making process. The slight difference in top gas composition is because industrial data contain a slight amount of O 2 and CH 4 in top gas, which does not exist in this model.

Effect of H 2 Injection on Coke Consumption Rate
The effect of injecting H 2 -rich gas to hearth and/or shaft on coke rate is shown in Figure 6. The highest H 2 injection rate is limited to 160 m 3 /tHM to ensure a balanced energy distribution in the BF shaft. Injecting H 2 to the BF hearth shows less coke consumption than injecting it to the shaft. This is because it provides more sensible heat to the lower part of the furnace to compensate the heat supplied by coke combustion. The relationship between H 2 injection and coke consumption rate at 9% oxygen enrichment rate is given as in Equation (55). Injecting 7.9~10.9 m 3 /tHM H 2 can reduce the coke consumption rate by 1 kg/tHM, depending on the injection position.
where X shaft and X hearth are the proportion of H 2 -rich gas injected to the shaft and hearth, respectively; and H 2 is the total H 2 injection volume to the BF, m 3 /tHM. The effect of H2 injection on RAFT at a constant oxygen enrichment rate is shown in Figure 6b. Injecting 20 m 3 /tHM H2 can reduce RAFT by around 16 °C. Increasing oxygen enrichment can achieve thermal compensation to maintain a stable RAFT and reduce coke consumption, as shown in Figure 7. Compared to the cases without thermal compensation, further reduction in coke consumption is obtained at 309 kg/tHM with 12.4% oxygen enrichment at 160 m 3 /tHM H2 injection. The effect of H 2 injection on RAFT at a constant oxygen enrichment rate is shown in Figure 6b. Injecting 20 m 3 /tHM H 2 can reduce RAFT by around 16 • C. Increasing oxygen enrichment can achieve thermal compensation to maintain a stable RAFT and reduce coke consumption, as shown in Figure 7. Compared to the cases without thermal compensation, further reduction in coke consumption is obtained at 309 kg/tHM with 12.4% oxygen enrichment at 160 m 3 /tHM H 2 injection. Figure 6. (a) Coke consumption rate at different H₂ injection volumes at different injection positions; (b) effect of H₂ injection to hearth on RAFT at 9% oxygen enrichment rate.
The effect of H₂ injection on RAFT at a constant oxygen enrichment rate is shown in Figure 6b. Injecting 20 m 3 /tHM H₂ can reduce RAFT by around 16 °C. Increasing oxygen enrichment can achieve thermal compensation to maintain a stable RAFT and reduce coke consumption, as shown in Figure 7. Compared to the cases without thermal compensation, further reduction in coke consumption is obtained at 309 kg/tHM with 12.4% oxygen enrichment at 160 m 3 /tHM H2 injection.

Figure 7.
Coke consumption rate at different H₂ injection volumes at different oxygen enrichment rates for thermal compensation. RAFT was maintained at 2096 °C in this test.

Effect on H₂ Utilisation Efficiency
In the case of injecting H₂ to hearth, bosh gas composition is shown in Figure 8a. As the H₂ injection rate increased from 0 to 160 m 3 /tHM at a constant CO injection rate, CO content in bosh gas drops due to less coke consumption. N2 content also decreased since less hot blast is required for carbon combustion. There was no H₂O in the bosh gas because H₂O that formed from the iron oxide reduction by H₂ reacted with coke to generate H₂ again at high temperature. The amount of top gas and its composition is described in Figure 8b. With more H₂ injection, the moisture content in the top gas increases slightly from 2.72% to 3.12% when the H₂ injection rate reaches from 0 to 100 m 3 /tHM. This is because H₂ utilisation efficiency significantly dropped from 72.6% to 22.9%, as shown in Figure 9a. CO content in top gas decreased because of less CO in the bosh gas

Effect on H 2 Utilisation Efficiency
In the case of injecting H 2 to hearth, bosh gas composition is shown in Figure 8a. As the H 2 injection rate increased from 0 to 160 m 3 /tHM at a constant CO injection rate, CO content in bosh gas drops due to less coke consumption. N 2 content also decreased since less hot blast is required for carbon combustion. There was no H 2 O in the bosh gas because H 2 O that formed from the iron oxide reduction by H 2 reacted with coke to generate H 2 again at high temperature. The amount of top gas and its composition is described in Figure 8b. With more H 2 injection, the moisture content in the top gas increases slightly from 2.72% to 3.12% when the H 2 injection rate reaches from 0 to 100 m 3 /tHM. This is because H 2 utilisation efficiency significantly dropped from 72.6% to 22.9%, as shown in Figure 9a. CO content in top gas decreased because of less CO in the bosh gas and increased CO utilisation efficiency. CO 2 concentration was reduced by less than 2% because there was a significant drop of CO in the bosh gas and increase in CO utilisation efficiency is very limited. The H 2 injection promotes CO utilisation with the presence of the water-gas shift reaction. However, the overall effect of the water-gas shift reaction was very limited across the whole BF. The comprehensive gas utilisation efficiency gradually decreased from 43.6% to 39.0% when H 2 injection reached 160 m 3 /tHM. H 2 and CO utilisation efficiency after thermal compensation is shown in Figure 9b. Results in this simulation reflect a similar trend as in the literature results [8,49]. Oxygen enrichment increased, and the nitrogen composition in the blast decreased with H 2 injection. Hence, there was more increase in reducing gas concentration than that in Figure 9a. With less N 2 dilution and stronger indirect reduction, CO utilisation with thermal compensation was enhanced by 5% with H 2 injection from 0 to 160 m 3 /tHM.
By H 2 increasing in the BF, CO utilisation increased, and H 2 utilisation efficiency decreased. One reason is that water-gas shift reaction Equation (18) would tend to proceed to the right-hand side, and both H 2 and CO 2 are generated in the upper furnace below 1000 • C. With the regeneration of H 2 , FeO reduction in Equation (7) can be considered to be proceeding in two successive stages, as shown in Equations (13) and (18) (7) : Overall : FeO (s) + CO (g) → Fe (s) + CO 2 (g) was very limited across the whole BF. The comprehensive gas utilisation efficiency gradually decreased from 43.6% to 39.0% when H2 injection reached 160 m 3 /tHM. H2 and CO utilisation efficiency after thermal compensation is shown in Figure 9b. Results in this simulation reflect a similar trend as in the literature results [8,49]. Oxygen enrichment increased, and the nitrogen composition in the blast decreased with H2 injection. Hence, there was more increase in reducing gas concentration than that in Figure  9a. With less N2 dilution and stronger indirect reduction, CO utilisation with thermal compensation was enhanced by 5% with H2 injection from 0 to 160 m 3 /tHM.  By H2 increasing in the BF, CO utilisation increased, and H2 utilisation efficiency decreased. One reason is that water-gas shift reaction Equation (18) would tend to proceed to the right-hand side, and both H2 and CO2 are generated in the upper furnace be- gradually decreased from 43.6% to 39.0% when H2 injection reached 160 m 3 /tHM. H2 and CO utilisation efficiency after thermal compensation is shown in Figure 9b. Results in this simulation reflect a similar trend as in the literature results [8,49]. Oxygen enrichment increased, and the nitrogen composition in the blast decreased with H2 injection. Hence, there was more increase in reducing gas concentration than that in Figure  9a. With less N2 dilution and stronger indirect reduction, CO utilisation with thermal compensation was enhanced by 5% with H2 injection from 0 to 160 m 3 /tHM.  By H2 increasing in the BF, CO utilisation increased, and H2 utilisation efficiency decreased. One reason is that water-gas shift reaction Equation (18) would tend to proceed to the right-hand side, and both H2 and CO2 are generated in the upper furnace be- Due to the smaller size and high diffusivity of H 2 , the reaction in Equation (13) has an advantage over the reaction in Equation (7). Thus, FeO reduction by CO was promoted by increased H 2 content.
As depicted in Figure 10, with H 2 injection and thermal compensation, the degree of indirect reduction increased because H 2 reduction replaced part of the direct reduction and oxygen enrichment enhanced reducing gas atmosphere. Since the direct reduction was a huge endothermic reaction process in the lower furnace, less direct reduction reduces coke consumption. Compared to the higher H 2 injection volume, injecting 0 to 80 m 3 /tHM H 2 generates more effect on the degree of indirect reduction, from 0.107 to 0.125. Further, the injection of more than 50 m 3 /tHM H 2 significantly increased the indirect reduction of CO. Considering gas utilisation efficiency, H 2 injection should be around 50-80 m 3 /tHM to maintain smooth BF operation and avoid too much excess H 2 in top gas. was a huge endothermic reaction process in the lower furnace, less direct reduction reduces coke consumption. Compared to the higher H2 injection volume, injecting 0 to 80 m 3 /tHM H2 generates more effect on the degree of indirect reduction, from 0.107 to 0.125. Further, the injection of more than 50 m 3 /tHM H2 significantly increased the indirect reduction of CO. Considering gas utilisation efficiency, H2 injection should be around 50-80 m 3 /tHM to maintain smooth BF operation and avoid too much excess H2 in top gas.

Effects on CO2 Emissions and Energy Consumption
In this work, the CO injected into the BF comes from the CO2 captured in top gas, which reduces CO2 emission compared to a traditional BF process. The emission from CCU can be negligible because it applies renewable electricity. The total emission of this iron-making process includes uncaptured CO2, flue gases from BFG combusted in the preheater and the hot blast stoves, which is shown in Figure 11a. The main CO2 emission reduction comes from the hot stoves. Compared with the traditional BF iron-making system that lacks CCU and gas injection, CO2 emission dropped from 534 to around 272 m 3 /tHM with 160 m 3 /tHM H2 injection in this system. When increasing H2 content from 0 to 80 m 3 /tHM, CO2 emission only decreased by 20 m 3 /tHM because more BFG was consumed to preheat the injection gas. Additionally, injecting H2-rich gas into the shaft showed more CO2 emission reduction capability than injecting H2 into a hearth or both tuyeres, as shown in Figure 11b.

Effects on CO 2 Emissions and Energy Consumption
In this work, the CO injected into the BF comes from the CO 2 captured in top gas, which reduces CO 2 emission compared to a traditional BF process. The emission from CCU can be negligible because it applies renewable electricity. The total emission of this ironmaking process includes uncaptured CO 2 , flue gases from BFG combusted in the preheater and the hot blast stoves, which is shown in Figure 11a. The main CO 2 emission reduction comes from the hot stoves. Compared with the traditional BF iron-making system that lacks CCU and gas injection, CO 2 emission dropped from 534 to around 272 m 3 /tHM with 160 m 3 /tHM H 2 injection in this system. When increasing H 2 content from 0 to 80 m 3 /tHM, CO 2 emission only decreased by 20 m 3 /tHM because more BFG was consumed to preheat the injection gas. Additionally, injecting H 2 -rich gas into the shaft showed more CO 2 emission reduction capability than injecting H 2 into a hearth or both tuyeres, as shown in Figure 11b. The energy consumption of the process described in Figure 1 was calculated on the basis of static mass and energy balance, using Equation (56). The standard coal coefficient for each substance in a kilogram of coal equivalent per ton of hot metal (kgce/tHM) is used in this analysis [50]. The energy consumption of the process described in Figure 1 was calculated on the basis of static mass and energy balance, using Equation (56). The standard coal coefficient for each substance in a kilogram of coal equivalent per ton of hot metal (kgce/tHM) is used in this analysis [50].
where E net is the net energy consumption of the process; E coke is the energy input by coke; E PC is the energy input by pulverised coal; E BOFG is the energy input by BOFG to the hot blast stoves; E blast is the energy carried by the blast to hot blast stoves; E cap is the electricity required for CO 2 capture unit; E conv is the electricity required for CO 2 electrolyser, as calculated by Equation (2); E water is the energy carried by the make-up water required at the humidifier for H 2 generation in the electrolsyer; E export is the energy carried by the BFG that is exported to other processes in the integrated steel mill; and E O2 is the energy carried by the oxygen that is generated in the electrolyser and exported to other processes in the integrated steel mill. Energy consumption results are shown in Figure 12. In a traditional BF, carbon resources from coke and pulverised coal injection provide the primary energy input, which accounts for 80% in total. When hydrogen is injected into the BF from 0 to 160 m 3 /tHM, carbon only accounts for 65% to 50% of the total energy consumption. With H 2 -rich gas injection from 0 to 160 m 3 /tHM, net energy consumption increased from 541 to 698 kgce/tHM, which is higher than that of the traditional BF (504 kgce/tHM) because electricity required for CO 2 conversion to generate H 2 kept increasing from 158 to 369 kgce/tHM.  Energy consumption for H2-rich gas injection BF and traditional BF process (energy carried by the make-up water needed for the CO2 conversion and by BOFG was less than 1% of total energy consumption and thus not indicated in this figure).
In summary, increasing hydrogen injection volume can reduce coke consumption and CO2 emissions. For coke consumption and CO2 emission reduction, the hydrogen injection amount should be as much as possible, as long as it satisfies the energy balance in the BF. In this case, the hydrogen injection amount should be 160 m 3 /tHM. However, injecting too much H2 significantly reduces its utilisation efficiency and increases the net energy consumption of this process. Further study is recommended to develop a multiobjective optimisation model to balance these effects of hydrogen injection. In general, injecting H2 at around 80 m 3 /tHM may consume less energy and suppress CO2 emissions under the simulation conditions. Energy consumption to produce 1 tonne of hot metal in the case of injecting 80 m 3 /tHM H2 with 200 m 3 /tHM CO is indicated in Figure 13. Compared to the traditional BF as indicated by Table 2, coke consumption decreased by 43 kg/tHM. CO2 emission dropped from 534 m 3 /tHM for a traditional BF to 278 m 3 /tHM in this case (including gas heating flue gas, capture unit CO2 letdown, and stack flue gas). However, electricity consumption in the CO2 capture unit and electrolyser is one of the largest energy inputs. The economic impact of this CCU technology as an auxiliary system to the BF is highly Energy consumption for H 2 -rich gas injection BF and traditional BF process (energy carried by the make-up water needed for the CO 2 conversion and by BOFG was less than 1% of total energy consumption and thus not indicated in this figure).
In summary, increasing hydrogen injection volume can reduce coke consumption and CO 2 emissions. For coke consumption and CO 2 emission reduction, the hydrogen injection amount should be as much as possible, as long as it satisfies the energy balance in the BF. In this case, the hydrogen injection amount should be 160 m 3 /tHM. However, injecting too much H 2 significantly reduces its utilisation efficiency and increases the net energy consumption of this process. Further study is recommended to develop a multiobjective optimisation model to balance these effects of hydrogen injection. In general, injecting H 2 at around 80 m 3 /tHM may consume less energy and suppress CO 2 emissions under the simulation conditions. Energy consumption to produce 1 tonne of hot metal in the case of injecting 80 m 3 /tHM H 2 with 200 m 3 /tHM CO is indicated in Figure 13. Compared to the traditional BF as