Parameter Optimization for Hydrogen-Induced Fluidized Bed Reduction of Magnetite Iron Ore Fines

Abstract

Hydrogen-based direct reduced iron (HyDRI) produced by fluidized bed has great potential for achieving the target of net-zero carbon in steelmaking. However, when magnetite ores were used as feedstock, several process parameters showed influences on its fluidization and reduction behaviors. To confirm the dominant influencing factors and its optimum process condition, the orthogonal experimental method was conducted in the present study. The result shows that the primary and secondary influencing factors are oxidation temperature, oxidation content, MgO addition amount, and gas velocity. The optimum condition is that the magnetite iron ore is deeply oxidized at 800 °C, mixed with 1.5 wt.% of MgO powder, and reduced in the fluidized bed at a gas velocity of 0.45 m/s.

1. Introduction

Steel is the world’s most important engineering and construction material [1]. The total production of crude steel in 2020 was 1.95 billion tons, and the top three steel-producing countries were China, India, and Japan, respectively [2]. With the development of developing countries, the steel demand will further increase [3,4]. It is estimated that the global steel demand will grow by more than 60% by 2050 [5]. Iron and steel production is a CO₂ emissions-intensive sector, which was responsible for 2.6 Gt CO₂ emission (about 7% of global anthropogenic CO₂ emissions) in 2019 [6]. The two dominant steelmaking processes are the blast furnace–basic oxygen furnace (BF–BOF) route and the scrap-based electric arc furnace (EAF) route [7,8]. In terms of CO₂ emission, the BF–BOF route emits 1.6–2.2 tons CO₂ per ton of crude steel, which is more than twice the amount CO₂ emission than that of the EAF route (0.6 tons CO₂ per ton of crude steel) [9,10,11]. To achieve the goal of CO₂ neutrality within the process, the shift from BF–BOF to EAF is a trend for future steelmaking [8]. Due to the limited access to high-quality scrap, it can be partly replaced by direct reduced iron (DRI). The ongoing development of the hydrogen-based direct reduced iron (HyDRI)—EAF route has a high potential for reducing CO₂ emission. When EAF is operated with pure HyDRI, CO₂ emission can be reduced to 25–53 kg per ton of crude steel [12,13]. Therefore, how to produce HyDRI properly and efficiently becomes an interesting topic.

Many commercial processes use shaft furnaces or fluidized bed reactors as iron ore reduction units [14,15]. Fluidized beds usually show better reduction efficiency because of faster mass and heat transfer. An ongoing hydrogen-based-fine-ore reduction process, namely HYFOR, developed by Primetals Technologies, is one of the successful examples [16]. However, if the de-fluidization phenomenon occurs, i.e., the iron ore fines cannot be fluidized by the gas flow, the reduction efficiency would drop dramatically, and process control of the system would become impossible. In a hydrogen-induced fluidized bed, the newly formed metallic iron on the particle surface is sticky and has a high probability of forming iron aggregates [17]. To prevent de-fluidization, the particle surface can be modified by coating treatment [18]. One practical method is to mix MgO with the iron ore fines before charging into the fluidized bed reactor. The MgO acts as a physical barrier to reduce the frequency of contact between particles [19]. When the reduction temperature is higher than 900 °C, MgO reacts with Fe₂O₃/FeO and forms Fe₂MgO₄/FeO·MgO on the particle surface, which serves as a chemical barrier [20].

Hydrogen reduction of iron oxide is an endothermic reaction that requires external heat input into the system such as preheating the iron ore. When magnetite iron ore is used, it becomes oxidized in a preheat process. Based on our previous study, the raw magnetite iron ore could not be completely fluidized at the temperature range of 600–800 °C. The generated fresh metallic iron accumulated on the surface of the magnetite particle and an iron shell was formed. While an oxidation treatment of magnetite iron ore improved the fluidization behaviors. The pre-oxidation treatment promoted the formation of a porous structure. The porous structure improved the reducibility and also avoids the formation of iron shell on the particle surface, thus decreasing the de-fluidization tendency. However, additional MgO was still needed when the reduction temperature was higher than 650 °C [21]. It was also found that the oxidation temperature and oxidation content (deeply oxidized or party oxidized) of magnetite influenced the fluidization and reduction behaviors [22]. Higher oxidation temperature shows a better fluidization improvement effect but leads to a lower reduction rate in the later reduction stage. According to kinetic analysis, the diffusion of the iron ions was the rate-limiting step. A lower pre-oxidation temperature could improve the diffusion of the iron ions.

The current study determines the proper operating parameters through an orthogonal experimental method. Furthermore, the dominant factors that influence the fluidization and reduction behaviors of the magnetite iron ore are confirmed. Based on the results, practical guidance can be provided for pilot or industrial trials.

2. Experimental

2.1. Materials

A low raw-grade magnetite iron ore was used as raw material, and the chemical composition is shown in

Table 1. Chemical composition of the raw magnetite iron ore, wt.%.

1 Fetot	FeO	SiO2	Al2O3	CaO	MgO	MnO	P
59.50	19.66	7.05	1.13	3.03	2.00	0.17	0.7

¹ Fe_tot—Total iron content.

. The high-purity MgO powder (>99.5 wt.% MgO, size below 44 µm) was used as an anti-sticking additive. The particle size of the raw magnetite ore is in the range of 125–500 μm, which is achieved by mixing 50 wt.% of 125–250 μm and 50 wt.% of 250–500 μm. For the oxidation treatment, 1000 g of the raw magnetite ore was charged into a steel vessel and put into a conventional heat treatment furnace at 800, 900, and 1000 °C for a specific time. To endure a uniform oxidation of magnetite particles within the material layer, the material was stirred manually during the oxidation. Two types of oxidized material, namely partly oxidized material and deeply oxidized material were obtained at each oxidation temperature. It should be noted that oxidation degrees of the deeply oxidized material and partly oxidized material range from 94% to 97% and 50% to 65%, respectively. The oxidation degree was confirmed by the actual weight gain and theoretical weight gain. The theoretical weight gain can be calculated according to FeO content from the chemical analysis of the raw magnetite iron ore. The detailed calculation has been described in detail elsewhere [23].

2.2. Apparatus and Methods

The main apparatus in this study, as shown in Figure 1, was a fluidized bed reactor with a 68 mm inner diameter [21,24]. The principle of the experimental apparatus was to measure the weight change and differential pressure drop continuously during the reduction process. The factors affecting the fluidized reduction were selected as oxidation temperature (Factor A), oxidation content (Factor B), MgO addition amount (Factor C), and gas velocity (Factor D). The orthogonal experimental plan was designed using IBM SPSS software (Statistics 26, IBM, Armonk, NY, USA), where three levels of each factor were taken. The reduction temperature and H₂ content were 700 °C and 15.9 Nl/min, respectively, for all the experiments. The gas velocity was controlled by changing N₂ content. A sample of 400 g of the material was taken for each experiment. When the reduction degree reached 95% or the reaction time reached 90 min, the reducing gas was changed to N₂ for cooling. The detailed experimental programs are given in

Table 2. Orthogonal experimental factors and levels.

Factor	A Oxidation Temperature, °C			B Oxidation Content			C MgO Amount, wt.%			D Gas Velocity, m/s
Level	A1	A2	A3	B1	B2	B3	C1	C2	C3	D1	D2	D3
	800	900	1000	Raw	Partly	Deeply	0.5	1.0	1.5	0.35	0.40	0.45

and

Table 3. Orthogonal experimental plan.

No.	A	B	C	D
1	A3	B2	C3	D1
2	A3	B3	C1	D2
3	A2	B1	C3	D2
4	A2	B3	C2	D1
5	A2	B2	C1	D3
6	A1	B3	C3	D3
7	A1	B1	C1	D1
8	A3	B1	C2	D3
9	A1	B2	C2	D2

.

2.3. Definition of Reduction Degree (RD) and Average De-Fluidization Index (Ave.DFI)

The RD was defined as the ratio of removed oxygen to the total oxygen bonded to iron, where the removed oxygen was measured by the weight loss. The RD can be calculated based on Equations (1)–(3) [21,25,26].

$$RD=\left(1-\frac{O_{bounded to Fe at t=ti}}{O_{ bounded to Fe at t=t0}}\right)\times 100\%$$

(1)

$$O_{ bounded to Fe at t=ti}=\frac{O_{bounded to F{e}_2{O}_3}+O_{bounded to FeO}-\Delta m}{M_O}$$

(2)

$$O_{ bounded to Fe at t=t0}=1.5\times \frac{m_0\times F{e}_{tot}}{M_{Fe}}$$

(3)

where $O_{ bounded to Fe at t=t0}$ and $O_{bounded to Fe at t=ti}$ represent the amounts of oxygen that are bounded to iron at the start and during the reduction, $\Delta m$ and $m_0$ are the weight loss due to the loss of oxygen and the mass of input material, and $M_O$ and $M_{Fe}$ are the constant molar masses for oxygen and iron.

A de-fluidization index (DFI) was introduced previously, as expressed by Equations (4) and (5) [22]. The DFI was a real-time value, which represented the portion of material that was not fluidized. For an easier comparison among the experiments in this study, an Ave.DFI was defined as shown in Equation (6). The differential pressure data were collected every two seconds. Dividing the total DFI by the amount of data gives Ave.DFI.

$$DFI=\frac{\Delta p_{calculated-bed}-\Delta p_{measured-bed}}{\Delta p_{calculated-bed}-\Delta p_{Fixed-bed}}$$

(4)

$$\Delta p_{calculated-bed}=\frac{m_t\times g}{Reactor area}$$

(5)

$$Ave.DFI=\frac{\sum DFI}{N_{DFI}}$$

(6)

where $\Delta p_{Fixed-bed}$ is the differential pressure drop when the material is in a fixed bed state. $\Delta p_{Fixed-bed}$ is 2 mbar in this work; $m_t$ is the mass of the remaining material in the fluidized bed during the reduction. $N_{DFI}$ is the amount of DFI data.

3. Results and Discussions

3.1. Orthogonal Experiment Analysis

The goal of a fluidized bed reduction is to obtain a high reduction rate at a stable fluidization state. Therefore, the time to reach RD = 90% (t_90%) and the Ave.DFI were chosen as indicators for the optimization of reduction parameters. The result is shown in

Table 4. The orthogonal experimental result.

No.	t90%, min	Ave.DFI, %
1	50.80	2.32
2	70.92	1.73
3	56.64	2.89
4	47.73	16.81
5	49.86	16.95
6	38.62	4.23
7	56.92	18.52
8	66.24	4.56
9	46.95	2.80

. Considering the fluctuation of the collected differential pressure data, it can be considered as a completely fluidized bed when the Ave.DFI is smaller than 5% [22]. Experiment No.2 shows the lowest Ave.DFI, and experiment No.6 shows the smallest t_90%. The orthogonal experiment analysis is conducted to confirm the dominant factors that influence the fluidization and reduction behaviors and to figure out the optimum condition. The results are shown In

Table 5. The orthogonal experimental analysis regarding t_90%.

Factor	A	B	C	D
K1	142.49	179.80	177.70	155.45
K2	154.23	147.61	160.92	174.51
K3	187.96	157.27	146.06	154.72
k1	47.50	59.93	59.23	51.82
k2	51.41	49.20	53.64	58.17
k3	62.65	52.42	48.69	51.57
R	15.16	10.73	10.55	6.60
Primary and secondary factors			ABCD
Optimization scheme			A1B2C3D3

and

Table 6. The orthogonal experimental analysis regarding Ave.DFI.

Factor	A	B	C	D
K1	25.55	25.97	37.20	37.65
K2	36.65	22.07	24.17	7.42
K3	8.61	22.77	9.44	25.74
k1	8.52	8.66	12.40	12.55
k2	12.22	7.36	8.06	2.47
k3	2.87	7.59	3.15	8.58
R	9.35	1.30	9.25	10.08
Primary and secondary factors			DACB
Optimization scheme			D2A3C3B2

. This analysis method is also conducted by Zhang et al. [27] and Xu et al. [28] to determine the optimum fluidized conditions during the reduction of hematite using a CO and CO-H₂ mixture.

The K_i is the sum of the experiment result (see Table 4) for the corresponding level number I as given in Table 3; k_i equals K_i divided by n, where n is the number occurrences of each level and n = 3; R = max{k_i} − min{k_i}. The experimental factor with the highest R-value represents the most dominant influencing factor. The experimental level with the smallest k_i value indicates the optimum choice in the corresponding experimental factor. Therefore, regarding reduction efficiency, i.e., t_90%, the primary and secondary influencing factors are oxidation temperature, oxidation content, MgO addition amount, and gas velocity. The optimum condition should be A₁B₂C₃D₃: the magnetite iron ore is partly oxidized at 800 °C, mixed with 1.5 wt.% of MgO, and reduced in the fluidized bed at a gas velocity of 0.45 m/s. Regarding the fluidization behaviors, i.e., Ave.DFI, the primary and secondary influencing factors are gas velocity, oxidation temperature, MgO addition amount, and oxidation content. The optimum condition should be D₂A₃C₃B₂: the magnetite iron ore is partly oxidized at 1000 °C, mixed with 1.5 wt.% of MgO, and reduced in the fluidized bed at a gas velocity of 0.4 m/s. From the orthogonal experimental result given in Table 4, fluidization is not a significant concern under experimental conditions. Experiment No.6 shows the fastest reduction rate. Only experiments No. 4, 5, and 7 are partly de-fluidized. The following discussions mainly concentrate on the reduction rate.

If the condition A₁B₂C₃D₃ shows the lowest t_90% with an acceptable Ave.DFI (<5.0%), it can be confirmed as the optimum condition. Thereafter, experiment No.10 is carried out using the condition A₁B₂C₃D_3. The t_90% and Ave.DFI of No.10 are 42.1 min and 2.4%, respectively. The reduction and de-fluidization curves are shown in Figure 2. Experiment No.10 shows a good reduction rate, but still, less than that of experiment No.6. The fluidization behaviors of No.10 and No.6 are similar. Thus, it can be concluded that the experiment. No.6 is the optimum condition instead of No.10.

3.2. The Reduction Curve Analysis

To further understand the orthogonal experimental result, a detailed reduction curve analysis is required. As discussed in Section 3.1, the oxidation temperature and oxidation content are the first two dominant influencing factors. Hence, Figure 3 shows the comparison reduction curve between the samples with different oxidation treatments, where the MgO amount and gas velocity are varied. In Figure 3a, it is shown that the deeply oxidized materials with a higher oxidation temperature give a lower reduction rate. The gaps between the reduction curves are bigger, especially when the reduction degree reaches 80%. As for the raw magnetite sample, the reduction curve presents a different shape, where the reduction rate is not restricted significantly in the later reduction stage. The partly oxidized materials show the same trend that a higher oxidation temperature gives a lower reduction rate (see Figure 3b), whereas the later reduction stages are improved compared with the deeply oxidized materials. This phenomenon is more obvious in the sample oxidized at 1000 °C, as shown in Figure 3c. The samples oxidized at 800 °C are an exception, as it is shown in Figure 2a that the deeply oxidized sample still shows a higher reduction rate than the partly oxidized sample even after reaching the target reduction degree i.e., >90%. For the raw magnetite sample, comparing tests No.8 and No.7, (see Figure 4), it is interesting to note that the test with a higher MgO addition amount and a higher velocity shows a lower reduction rate. It gives a hint that when using magnetite iron ore as raw material, a critical velocity must be determined. It is not that the higher the gas velocity, the faster the reduction rate.

3.3. Structural Analysis

To figure out the reason for the reduction behaviors, the morphology of the reduced samples was analyzed via an optical microscope. The reduced particles of the raw magnetite sample are shown in Figure 5. According to Hayes et al. [29,30], the iron morphology obtained by reducing magnetite is mainly controlled by the reduction mechanism. Wolfinger et al. [31] share a similar opinion and found that, for magnetite iron ore ultra-fines, when the reduction is conducted at low temperatures (<675 °C) and controlled by the chemical reaction in the initial reduction stage, a porous iron layer can be formed. However, at higher temperatures (>750 °C), a dense iron layer is formed due to the fast generation of iron and built up in the wüstite. In the present study, the surfaces of the reduced magnetite particles are covered by porous iron layers. From Figure 5a,d, a dense wüstite core is observed, and the porous iron layer is thinner than the other two reduced magnetite samples. It indicates that the reduction of the raw magnetite sample is controlled by the chemical reaction in the initial stage. As the reaction continues, the newly formed iron nuclei accumulate and build a dense iron core within the particles. Iron nuclei accumulate more rapidly at higher gas velocities and form a thinner, porous iron shell, leading to difficulties in gas diffusion within the particles. As shown in Figure 5b,c,e,f, the porous iron shells are much bigger than those in Figure 5a,d. The gas diffuses more easily into the particles, resulting in a higher reduction rate.

As shown in Figure 6a,b,d,e, for the oxidized samples oxidized at 800 °C, the obtained iron morphology is different from that of raw magnetite samples. Instead of porous iron layers, many coarse iron grains are observed on the surface of the particles. From Figure 6c,f, it is seen that, in the sample with oxidation treatment at 1000 °C, the inner part of the reduced particle is porous as well. However, many wüstite islands are observed, indicating a poor reduction behavior. The iron morphology is consistent with our previous study [22]. The oxidation treatment parameters, including oxidation temperature and oxidation content, lead to changes in the reduction kinetic mechanism. The changes in kinetic mechanism are due to the coarse porous wüstite obtained during the reduction of oxidized magnetite [31]. The target of the present study is to confirm the optimum condition for reduction of magnetite ore. The polished section images of reduced samples support the conclusion drawn from the reduction curves, as discussed in Section 3.2.

4. Conclusions

In conclusion, the optimum condition for hydrogen-induced fluidized bed reduction of magnetite iron ore fines is successfully determined by an orthogonal experimental method. Furthermore, the dominant factors that influence the reduction behaviors of the magnetite iron ore are confirmed. Under the experiment’s conditions, the fluidization behavior is not a problem. Regarding reduction efficiency, the primary and secondary influencing factors are oxidation temperature, oxidation content, MgO addition amount, and gas velocity. The optimum condition is that the magnetite iron ore is deeply oxidized at 800 °C, mixed with 1.5 wt.% of MgO, and reduced in the fluidized bed at a gas velocity of 0.45 m/s.