Impacts of Blending Semi-coke in PCI Coal on Grinding Efficiency and Blast Furnace Operation

Abstract

Semi-coke is a product of low-temperature pyrolysis by low-rank coal, with a composition similar to that of anthracite for pulverized coal injection (PCI). Herein, we investigated the differences in grindability and combustibility between semi-coke and anthracite, analyzed the compositional and microstructural characteristics related to the performance of semi-coke, and assessed the impacts on grinding efficiency and blast furnace operation after replacing anthracite for injection with two types of semi-coke. Semi-coke is rich in high-hardness quartz that is tightly bound to the carbon matrix, making the semi-coke particles very hard, with a high Hardgrove grindability index (HGI) and high abrasion index. The addition of semi-coke reduced the grinding efficiency of the mill and afforded large-sized milled particles. The developed pore structure of semi-coke can enhance kinetic diffusion, and semi-coke is less ordered than coal, thereby providing more reactive sites for combustion reactions. These two reasons cause the ignition temperature of semi-coke to be significantly lower than that of coal. The addition of semi-coke increased the PCI ratio, decreased the fuel ratio, improved the permeability of the blast furnace, decreased the sulfur content in pig iron and carbon content in blast furnace dust. The difference in grinding productivity between semi-coke and coal widens as grinding time increases, suggesting that the HGI method may overestimate the actual grindability of semi-coke. The feasibility of reducing the grinding energy by optimizing particle sizes of semi-coke and improving the grindability of semi-coke by using selected pyrolytic coal and adjusting the pyrolysis temperature was proposed.

1. Introduction

Low-rank coal accounts for a relatively large proportion of total global coal reserves.¹⁾ The cascade utilization of low-rank coal with low-temperature pyrolysis as the core is one of the best routes for clean and efficient utilization of low-rank coal.^2,3,4) Semi-coke is the solid product of low-rank coal after low-temperature pyrolysis to remove some volatiles and separate high-value-added tar. In China, with the growing scale of the low-temperature pyrolysis industry, the production of semi-coke is increasing annually, and its capacity reached 120 million tons in 2021. Recently, semi-coke has been successfully used as a reducing agent in calcium carbide and ferroalloy production processes, but the use of semi-coke in these areas is still very limited, with nearly half of the semi-coke not being utilized each year.

Pulverized coal injection (PCI) is an important means to save coke and reduce smelting costs for blast furnace. As the pig iron production remains high and the PCI ratio continues to grow, the demand for high-quality injection fuels from steel companies is growing, and it is imperative to find new low-cost, high-quality injection fuels. Injection materials such as waste plastics, biomass charcoal, and waste tires have been used in some blast furnace injection processes;^5,6,7,8) however, their large-scale applications rely heavily on the easy access to these resources. The proximate analysis and heat output of semi-coke are close to that of the scarce anthracite for conventional injection, but its price is much lower than that of the latter; thus, >10 Chinese steel companies have conducted preliminary trials of semi-coke for blast furnace injection. Their results show that semi-coke has the advantages of good conveying performance, high safety, and high heat output.⁴⁾ However, during the experiment, some companies found that semi-coke adversely affects pulverized coal production and blast furnace operation, but no systematic research results show any clear correlation between the reason behind these problems and the microstructure of semi-coke.

Under the same pulverizing process and equipment conditions, the grinding efficiency is mainly influenced by the fuel properties and usually measured using the Hardgrove grindability index (HGI).^9,10,11) The HGI indicates how hard it will be to crush the material, which determines the energy consumption and grinding cost of material crushing.¹²⁾ At present, researchers have studied many factors influencing the grindability of coal and found out that the main influencing factors include coal rank, proximate analysis of coal, maceral of coal, and internal pore structure of coal.^13,14,15,16) The impacts of coal rank on grindability are that low-rank and high-rank coals have low grindability, and medium-volatile bituminous coal has high grindability.¹⁷⁾ As for proximate analysis, the components of the coal sample and HGI can be predicted by regression. With regard to the relationship between the maceral and grindability of coal, different brittleness and grindability between different macerals, vitrinite more easily broken into smaller particles.^18,19) As regards the relationship between the internal pore structure and HGI of coal, the stress analysis of the coal pore structure during the grinding process found that fracturing first occurs at the surface of pores.²⁰⁾ However, few researchers studied the factors influencing the HGI of semi-coke with a composition similar to coal, and how semi-coke mixed with coal will impact the grinding efficiency of the pulverizing system remains unclear.

Rapid combustion of fuel injected into the blast furnace raceway mainly provides heat and reducing gas for the blast furnace, and the heat output and combustion reactivity of the fuel are the factors limiting injection capacity.^21,22,23) Lower heat output will afford insufficient heat supply to the blast furnace, thus making it less efficient to substitute coke with coal; lower reactivity will increase the yield of unburned char at the end of the raceway, thus reducing the permeability/liquid permeability of the cohesive zone and dropping zone of the blast furnace.²⁴⁾ Therefore, it is hoped that the substitute fuel should have similar or higher heat output and combustion reactivity than the pulverized coal being replaced. The heat output of a fuel is most closely correlated to its fixed carbon content, and its intrinsic combustion reactivity is mainly influenced by its volatile content, particle size, pore structure, and carbon chemical structure under the same process conditions.^{25,26,27,28,29,30,31)} Therefore, research on these properties of semi-coke is essential to evaluate its impacts on the smelting process.

This study compared the grindability and combustion performance of semi-coke and PCI coal, assessed the impacts of partially replacing the blast furnace injection fuel with two types of semi-coke on the grinding efficiency and blast furnace conditions in a steel company, and discussed strategies to further improve the use of semi-coke as a blast furnace injection fuel.

2. Experimental

2.1. Samples

The samples of this study—semi-coke A (SA) and semi-coke B (SB)—were taken from a low-rank coal mid–low temperature pyrolysis enterprise in northwest China, and the pyrolysis reactor was an internally heated upright pyrolysis furnace with a production capacity of 75 kt/a. The pyrolysis process involved heating the raw coal using the gas produced by the combustion of large excess gas with a certain amount of air, and the pyrolysis furnace comprises a drying section, preheating section, pyrolysis section, and cooling section. The temperature of the pyrolysis section was 620°C–680°C, and the residence time of the material in the pyrolysis section was ~4.5 h. The raw material for pyrolysis was high volatile bituminous coal with a volatility of 34%–36%. The coal used for injection was low volatile bituminous coal CA and anthracite CB used in a 1880 m³ blast furnace of Shandong Iron and Steel Co., Ltd. of China, and the conventional ratio of the two was 6:4. The mixed coal was denoted as PC. The properties including moisture (M), ash (A), fixed carbon (FC), volatiles (V), sulfur (S) and calorific value (Q) of each fuel are shown in Table 1.

Table 1. Properties of the fuels used for experiments.

Fuels	Mad a/%	Ad b/%	FCd/%	Vd/%	St c/%	Qgr,d d/(MJ/kg)
CA	1.09	9.31	79.23	11.46	0.63	30.09
CB	1.16	10.75	81.02	8.23	0.56	30.21
PC	1.11	9.97	80.34	10.70	0.59	30.16
SA	1.26	11.67	80.15	8.18	0.37	29.01
SB	1.32	8.78	79.29	9.93	0.26	29.54

^a  ad indicates the abbreviation of ‘air dry basis’;

^b  d indicates the abbreviation of ‘dry basis’;

^c  indicates total sulfur content.

^d  indicates high calorific value of dry base.

The industrial injection experiment carried out in April to May 2020 and it comprised four stages: baseline stage I, test stage I, baseline stage II, and test stage II. The compositions of the injection fuels for the two baseline stages were that of the existing fuel, i.e., 60% CA + 40% CB, and each baseline stage lasted 10 days. In test stage I, CB was substituted by SA, which changed the fuel composition to 60% CA + 40% SA, and the stage lasted 10 days. In test stage II, CB was substituted by SB, which changed the fuel composition to 60% CA + 40% SB, and the stage lasted 10 days. The operating indicators of the grinding workshop and blast furnace were monitored during the experiment and compared with those in the baseline stage. The particle sizes of semi-coke used in the industrial injection experiment were <20 mm, and the semi-coke was fed into the medium-speed mill in proportion to the raw coal for grinding with an incoming hot flue gas temperature of ~320°C and gas consumption of 5500–6000 m³/h. The resulting mixed fuel from the mill came out from the coal injecting tank. During the experiment, the stability of other raw materials, fuels, and operations of the blast furnace, in addition to its injection fuel, were ensured.

In laboratory combustion tests, lumpy raw coal or semi-coke was first coarsely broken using a jaw crusher, then finely broken using a roller crusher, and finally ground to samples >74 μm in size using a ball mill.

2.2. Grinding Performance Characterization

HGI is used to evaluate how easily the fuel can be crushed, and the experiments were conducted as per the Chinese standard GB/T2565-2014. A 50 g air-dried coal sample with a particle size of 0.63–1.25 mm was put into the HGI tester. A 284 N force was applied to the steel ball to drive the motor to rotate at 60 rpm for grinding purposes. The pulverized coal was sieved on a sieve shaker with an aperture size of 71 μm, and the pulverized coal above and under the sieve was weighed. Finally, the calibration chart was checked to derive the HGI values. The higher the value, the lower the grinding performance of the fuel.

How the fuel abrades the coal grinding mill is characterized by the abrasion index (AI), and the experiments were conducted as per the Chinese standard GBT 15458-2006. A 2 kg coal sample with a particle size <9.5 mm (the constituents of coal <1.25 mm did not exceed 30%) was placed in the grinding tank equipped with four steel blades, and when the blades rotated at 12000 rpm under the specified conditions, the AI of the fuel was calculated based on the mass loss of the blades during rotation. The larger the value of mass loss, the higher its AI for the mill.

2.3. Combustion Performance Characterization

The combustion processes and gas escape characteristics of coal and semi-coke were tested using simultaneous thermal analysis and mass spectrometry (TA-MS), where mass and heat flux variation data were collected using a simultaneous thermal analyzer (Setsysevo, Setaram) and gas escape data were collected using a quadrupole mass spectrometer (ThermoStar, PfeifferVacuum). The test sample weighed 10 mg; the reactant gas was air; the flow rate was 50 ml/min; the heating rate was 15°C/min, the test temperature was 50°C–950°C; the generated gas was put into mass spectrometer via capillary tubes; and the data were recorded dynamically throughout the test. The fuel ignition temperature was analyzed using the TG-DTG method which is a common method for evaluating combustion performance.³²⁾

2.4. Characterization of Other Properties

A laser particle size analyzer (Rise-2002, RunZhi) was used to characterize the particle size distribution of the samples. The specific surface areas and pore structures of the samples were tested using a specific surface meter and pore size analyzer (JW-BK222, JingWei), respectively. The adsorption medium was high-purity N₂; the specific surface area and total pore volume of semi-coke were calculated using the BET method; and the pore volume was calculated using the BJH method. Prior to sample analysis, the sample was degassed in He gas flow at 120°C for 2.5 h. Surface topography and element distribution were measured using field emission scanning electron microscopy (SEM, GeminiSEM300, Zeiss) and energy dispersive spectrometry (EDS).

X-ray diffractometry (D8 Advance, Bruker) was used to analyze the inorganic mineral composition and microcrystalline structural characteristics of the fuel under the following conditions: voltage: 40 kV, current: 30 mA, Cu target, scanning range: 5°–90°, scanning speed: 3°/min. The microcrystalline structure was characterized by the parameters d₀₀₂, L_c, and L_a, which were calculated using Scherrer’s formula and Bragg’s equation.³³⁾   

$$d_{002}=\frac{\lambda }{(2sin{\theta }_{002})}$$(1)

  

$$L_c=\frac{k_1\lambda }{({\beta }_{002}cos{\theta }_{002})}$$(2)

where d₀₀₂ is the distance between the individual aromatic layers of the sample; L_c is the height of the microcrystal perpendicular to the aromatic layer sheets; β₀₀₂ is the half-height width of the strongest diffraction peak; λ is the wavelength of the incident X-rays (0.15406 nm); and k₁ is the waveform factor (0.89).

3. Results

3.1. Properties of Semi-coke Related to Grinding Efficiency

The HGIs and AIs of various fuels are shown in Fig. 1. The HGIs of the two PCI coals CA and CB are 79 and 59, respectively, and that of the mixed coal PC is 69, which is consistent with the law of weighting. The HGIs of the two semi-coke samples, SA and SB, are only 49 and 44, respectively, indicating that the grindability of semi-coke is significantly lower than that of PCI coal, which may reduce the grinding efficiency if a large proportion of semi-coke is added. Meanwhile, hard-to-grind substances cause abrasion on the grinder rolls and pipes, the magnitude of which can be evaluated by the AI. It was found that the AIs of CA and CB were 29.25 and 32.21 mg/kg, respectively; the AI of mixed coal PC was 30.63 mg/kg. The AIs of SA and SB were as high as 62.74 and 77.55 mg/kg, respectively, which were about twice that of PCI coal, indicating that the abrasion ability of semi-coke was significantly higher than that of PCI coal. The abrasion ability of SB was stronger than that of SA. The above results show that compared with PCI coal, semi-coke itself possesses low grindability and may adversely affect the coal grinding efficiency when used alone. The abrasion caused by semi-coke on the grinding mill may be higher, and we should pay attention to the service life of the mill parts when semi-coke is actually added.

Fig. 1.

HGIs and AIs of the samples. (Online version in color.)

The reasons for the low HGIs of semi-coke samples were analyzed from the perspectives of proximate analysis, hardness, and mineral composition, which affect grindability. As shown in Table 1, the moisture, fixed carbon, and volatile content of semi-coke are close to those of PCI coal, and therefore should not be the main influencing factors. Hardness is an important performance indicator to measure the hardness of a material, which can be understood as the ability of a material to resist elastic deformation, plastic deformation, or destructive force. The higher the hardness of a material, the lower its grindability; thus, the hardness value is often used as one of the important indicators to measure the grindability of a material. Table 2 presents the microhardness and Mohs hardness values of PCI coal and semi-coke, and the comparison shows that the microhardness of semi-coke is 3.3–6.7 times that of PCI coal, and its Mohs hardness is 1.6–1.9 times that of PCI coal. It can be assumed that the important reason for the significant difference in grindability between the two is that semi-coke has higher hardness. The ash composition of semi-coke and PCI coal also differs significantly. PCI coal has the highest content of Al₂O₃ and SiO₂, and therefore high content of clay minerals, which are of medium hardness and have an impact on the grindability higher than that of quartz and lower than that of carbonate minerals. Semi-coke has a higher content of SiO₂ and CaO and lower content of Al₂O₃, corresponding to the top three minerals of quartz, carbonate, and clay minerals, respectively. The phase composition of the inorganic minerals in the fuel was further tested using XRD, as shown in Fig. 2. PCI coal ash contains minerals such as kaolinite, mullite, calcium aluminum yellow feldspar, hard gypsum, hematite, and quicklime, while inorganic minerals in semi-coke ash are mostly quartz, kaolinite, gehlenite, and quicklime. As can be seen, the biggest difference between the inorganic minerals in the two fuels is that half of semi-coke is quartz minerals, which are hard inorganic minerals that are extremely difficult to crush and can significantly reduce the grindability of coal.³⁴⁾ Their distributions are discussed below.

Table 2. Fuel properties related to grindability.

Property		CA	CB	SA	SB
Hardness	Microhardness (kgf/mm2)	26	35	111.7	174.8
	Mohs hardness/(HM)	2.1	2.2	3.7	3.9
Ash composition	SiO2	30.43	37.47	49.5	54.07
	CaO	14.21	13.42	18.75	12.52
	Al2O3	34.59	31.53	13.32	17.9
	Fe2O3	3.21	2.85	7.69	6.43
Inorganic mineral composition by optical microanalysis	Clay minerals	56.20	66.94	10.34	10.64
	Anhydrite	33.21	29.80	8.62	8.94
	Carbonates	6.93	2.45	32.76	25.53
	Quartz	3.65	0.82	48.28	54.89

Fig. 2.

XRD results of the ash of semi-coke and PCI coal. (Online version in color.)

To determine the appropriate semi-coke ratio for industrial experiments, we investigated the impacts of different proportions of semi-coke on the HGIs and AIs of mixed fuels, and the results are shown in Fig. 3. As shown in Fig. 3(a), the HGI decreased with increasing addition proportions of both semi-coke samples, and the actual and theoretical weighted HGIs were basically the same when their addition proportions were <40%, indicating that there was basically no synergistic effect between a hard-to-grind substance and easy-to-grind one. The actual HGI value was found to be significantly lower than the theoretical value when the semi-coke proportion reached 60%, indicating that the grindability of the mixed fuel decreases with increasing hard-to-grind semi-coke content. Additionally, it can be found that the relatively hard-to-grind SB affects the HGI to a greater extent than the relatively easy-to-grind SA.

Fig. 3.

Impacts of different proportions of semi-coke on HGI and AI of the mixed fuel. (Online version in color.)

As shown in Fig. 3(b), as the AI increases with increasing proportion of two semi-coke samples, the impact of SB becomes greater than that of SA, indicating that a higher proportion of semi-coke may cause more serious abrasion on metal devices inside the mill and thereby shorten its maintenance cycle. It can also be found that the theoretical AI and actual weighted AI differ significantly at the proportions of 20% and 60%, where the theoretical AI is lower than the actual AI at the lower proportion but higher than the actual value at the higher proportion. This indicates that as the proportion of high-abrasion materials increases, the abrasion of the mixed sample on the mill will gradually change from being dominated by low-abrasion materials to being dominated by high-abrasion materials. Analysis of HGI and AI also showed that the grindability and abrasion of semi-coke mixed samples are strongly influenced by the properties of semi-coke itself, and more so with the increase of its proportion. In the actual grinding process, the proportion of fuels like semi-coke, which are both difficult to grind and easy to abrade the mill, should be strictly controlled. In the subsequent industrial experiment, CA was mixed with SA and SB, as the high grindability and low abrasion of CA can compensate for the low performance of semi-coke. As the coal/semi-coke mix with 40% semi-coke basically follows the law of weighting and the two indices of the mixed fuel and PC variation are acceptable, 40% was used as the proportion for industrial experiments.

3.2. Impacts of Adding Semi-coke on the Grinding Efficiency

The indicators of the grinding process during the industrial experiment are shown in Fig. 4. By comparing the grinding efficiencies of the mill in the test stage and baseline stage, we found that the average grinding efficiency decreased by 2.5 and 3.3 t/h after SA and SB were added to the raw coal, respectively, which increased the average pressure difference between the inlet and outlet of the mill by 486 and 578 Pa, respectively. The decrease in grinding efficiency and increase in differential pressure were caused by the decrease in grindability of the material being ground and decrease in crushing speed at the same coal feed rate, respectively, which led to fuel build-up. This indicates that semi-coke does adversely affect grinding efficiency, with SB having a higher impact than SA, which is consistent with the results of the HGI predictions for semi-coke.

Fig. 4.

Indicators for the grinding process in the industrial experiment. (Online version in color.)

The grinding capacity of the grinding workshop of the enterprise in this experiment exceeded the actual demand of the blast furnace, which has no significant impact on the actual production of the blast furnace, except for the increase in electricity required for grinding the same quality of fuel. Additionally, an increase of 20.5% in the amount of residue (mainly unground coal particles) at the mill outlet was observed during the test stage compared to the baseline stage, again indicating that the presence of more extremely-difficult-to-grind materials in the semi-coke reduced the grinding efficiency.

The particle size distribution and characteristic parameters of the products from the small-scale coal grinding tests and coal bunker samples are shown in Fig. 5 and Table 3, respectively. Figure 5(a) presents the particle size distribution of the sample under a 200-μm mesh sieve measured using a HGI tester. The particle size distributions of the two semi-coke samples shift toward the larger-size section, with SB shifting more. Figures 5(b) and 5(c) present the particle size distributions of the coal bunker samples in the mill after adding SA and SB, respectively, which are similar to those when the HGI tester is used, i.e., semi-coke causes the grinding products to shift toward the larger-size section. Table 3 presents the characteristic parameters of fuel particle size distribution after grinding. The parameters D₁₀, D₅₀, and D₉₀ were used to characterize the particle size distribution of the samples, that is, the particle diameter value at which the cumulative distribution percentage reached 10%, 50% and 90%, respectively. These parameters increase with semi-coke, with SB increasing more. The dispersion of particle size distribution, i.e., Span, of the powder was calculated using Eq. (3).³⁾   

$$Span=[D(v,90)-D(v,10)\left]/D(v,50)\right]$$(3)

Under laboratory grinding conditions, the difference in Span value between each semi-coke sample and PCI coal was small; however, under industrial grinding conditions, the Span values of semi-coke were significantly lower than those of PCI coal, indicating that semi-coke afforded a more concentrated particle size distribution of the grinding products. As the industrial grinding process limits the maximum particle size of the output powder, and D₁₀ with semi-coke added is higher than that of PCI coal, the Span of semi-coke decreased. The above results indicate that the low grindability of semi-coke affects the particle size distribution of the mixed samples.

Fig. 5.

Particle size distributions of pulverized fuels. (Online version in color.)

Table 3. Characteristic parameters of particle size distribution of pulverized fuels.

Sample		D10/μm	D50/μm	D90/μm	Span
Lab tests	PC	29.44	45.26	66.97	0.83
	SA	32.51	49.74	73.24	0.82
	SB	39.10	61.23	89.34	0.82
Grinding workshop	PC	47.22	85.58	135.96	1.04
	CA+40%SA	55.45	88.51	133.59	0.88
	CA+40%SB	55.79	95.41	139.58	0.88

3.3. Properties of Semi-coke Related to Combustion Performance

The combustion performance and heat output of the injection fuel at the front end of the air outlet are decisive factors limiting the increase in injection ratio and replacement ratio of a blast furnace. Thus, we compared and analyzed the combustion reactivity and heat output of semi-coke and PCI coal to obtain the TG-DTG-DSC-MS curves of the combustion processes of semi-coke and PCI coal using thermal analysis, the results of which are shown in Fig. 6. The combustion characteristic parameters were derived from the thermal analysis curves, the results of which are shown in Table 4. As shown in Figs. 6(a) and 6(b), compared with PCI coal, the combustion curves of both semi-coke samples shift toward the lower-temperature section. The ignition temperature is lowered by ~25°C, and the burnout temperature is lowered by ~15°C. As shown in Table 4, the ignition temperature of semi-coke is ~80°C lower than that of PCI coal. As shown in Fig. 6(c), the heat release rate of semi-coke during combustion is basically the same as that of PCI coal before 350°C, but higher than that of PCI coal after 350°C, and the heat release ends at a lower temperature. As shown in Table 4, the actual cumulative heat flux values obtained from the DSC curve integration for the four fuels do not differ significantly. As shown in Fig. 6(d), the shapes of the CO₂ gas release curves of the semi-coke and PCI coal combustion processes are similar, except that the gas release curve of semi-coke shifts to the lower-temperature section and the fastest release rate corresponds to a lower temperature. The above results show that the combustion reactivity of both semi-coke samples is higher than that of PCI coal, and the replacement of PCI coal with semi-coke will not adversely affect the fuel conversion efficiency in the raceway.

Fig. 6.

TG-DTG-DSC-MS curves of fuel combustion processes: (a) TG curve, (b) DTG curve, (c) DSC curve, and (d) CO₂ mass spectrometry curve. (Online version in color.)

Table 4. Characteristic parameters of fuel combustion processes.

Samples	Ignition temperature/°C	Burnout temperature/°C	Maximum reaction rate (%/min)	Temperature corresponding to fastest rate/°C	Actual heat flux of DSC combustion/ (μv·min/mg)	Temperature corresponding to the fastest CO2 release rate/°C
CA	500	661	−9.08	584	538	633
CB	514	721	−6.71	613	536	659
SA	429	619	−7.44	579	534	603
SB	431	651	−6.13	586	531	630

The combustion performance of a fuel is mainly influenced by factors such as volatile content, pore structure, carbon chemical structure, and mineral composition. As the differences in volatile and ash contents between semi-coke and PCI coal were not significant (Table 1), we mainly compared their differences in pore structure parameters and carbon chemistry, and the results are shown in Table 5. The specific surface area of semi-coke is >10 times that of PCI coal, indicating that semi-coke can provide more carbon–oxygen reaction interfaces during combustion; meanwhile, the pore volume of semi-coke is significantly larger than that of the injection fuel, which makes it easier for the reactant gases and gas products to diffuse in the pores of particles, thus reducing the impact of internal diffusion on the reaction. Unlike PCI coal, numerous micropores and mesopores are present in semi-coke, making the average pore size of semi-coke smaller than that of PCI coal. During the pyrolysis of low-rank coals, volatiles release gradually as temperature increases, and the original pores in the coal gradually expand and generate abundant new pores, thus developing the pore structure of semi-coke.³⁷⁾ As shown in the SEM images in Fig. 7, the surfaces of semi-coke particles are relatively rough, with more macroscopic pores and cracks, which also increase the specific surface area of semi-coke, create more interfaces for gas–solid reactions, and facilitate the diffusion of reactants and products, thus improving the combustion reactivity. With regard to the carbon structure, it can be seen from the carbon chemical structural parameters in Table 5 that the d₀₀₂ values of semi-coke and PCI coal are very close, and the L_c and directional layer size of semi-coke are lower than those of PCI coal. The L_c of semi-coke is >50% that of the raw coal. The lower aromatic layer stacking height and size of semi-coke indicate that it is less ordered than PCI coal. The rich defective microcrystalline structure of semi-coke provides more reactive sites for combustion reactions, which is favorable for the early ignition of semi-coke.³⁵⁾

Table 5. Fuel structure parameters related to combustion performance.

		CA	CB	SA	SB
N2 adsorption results	Specific surface area/(m2/g)	2.370	1.909	39.663	27.849
	Total pore volume/(cm3/g)	0.004	0.005	0.037	0.028
	Average pore size/(nm)	7.314	10.218	3.714	4.065
XRD results	d002/nm	0.360	0.359	0.363	0.366
	Lc/nm	2.812	3.458	1.299	1.364
	La/nm	3.977	4.115	3.477	3.673

Fig. 7.

Morphologies of PCI coal and semi-coke: (a) CA and (b) SA.

The ignition temperatures and heat outputs of the mixed fuels with different semi-coke additions are shown in Fig. 8. As shown in Fig. 8(a), with increasing semi-coke addition, the ignition temperature of the mixed fuel decreased significantly and was close to the theoretical weighted value before the semi-coke addition reached 40%. When the semi-coke addition was increased to 60%, the actual ignition temperature was lower than the theoretical one, indicating that at this point, the high combustion reactivity of semi-coke played a synergistic reinforcing role for the combustion of the fuel as a whole. As shown in Fig. 8(b), because the heat output of semi-coke is slightly lower than that of PCI coal, the total heat output of the mixed fuel showed a decreasing trend with increasing semi-coke addition, and the theoretical and actual values were basically the same. Since the heat output of SA is lower than that of SB, the total heat output decreased more significantly after SA was added.

Fig. 8.

Ignition temperature and heat output of mixed fuels with different semi-coke additions. (Online version in color.)

3.4. Impacts of Semi-coke Addition on Blast Furnace Operation

The blast furnace indicators before and after the addition of semi-coke in the injection fuel are shown in Fig. 9. The data points are recorded each day during the test. Compared with the baseline stage, the impacts of SA and SB additions in the test stage on the fuel consumption and blast operation indicators have the same trend, which is manifested in that: ① the coal injection ratio increased, coke ratio decreased, and fuel ratio decreased, which indicates that semi-coke can replace PCI coal in the furnace; the coal/coke replacement ratio also increased and fuel conversion efficiency improved; ② the decrease in carbon content in the blast furnace dust shows that there is no increase in unburned char caused by insufficient combustion and gasification in the raceway, and even the transformation of semi-coke in the furnace is more thorough than that of PCI coal; ③ the permeability index of the blast furnace increased, indicating that the conversion rate of semi-coke in the raceway is high; the amount and consumption of unburned char in the furnace basically maintain a dynamic balance, and less unburned char improved the permeability and liquid permeability of the cohesive zone; ④ the sulfur content in molten iron was significantly reduced as the sulfur content in semi-coke is lower than that in PCI coal (Table 1), which reduced the amount of sulfur in molten iron.

Fig. 9.

Blast furnace operation indicators for industrial experiments. (Online version in color.)

In summary, after 40% semi-coke was added to the blast furnace injection fuel, the furnace operation was stable; the fuel utilization improved compared with the baseline stage; and the impacts of SA and SB addition on the blast operation indicators were similar. This is mainly due to the fact that the combustion reactivity of semi-coke is higher than that of the anthracite coal being replaced and does not produce more unburned char at the end of the raceway zone. The most fundamental reason is the well-developed pore structure of semi-coke, which improves kinetic diffusion. Meanwhile, semi-coke is chemically less ordered than PCI coal, and therefore provides more reactive sites for combustion reactions.

4. Discussion

4.1. Analysis of the Reasons for the Decrease in Grinding Efficiency after Adding Semi-coke

The structurally dense and difficult to grind large-sized semi-coke particles obtained from the grinding bowl after long-term grinding during the small-scale coal grinding tests were characterized by SEM-EDS, as shown in Fig. 10. The surfaces of semi-coke particles are smooth, and the walls of larger-sized pores with weaker bonding have been worn away without macroscopic cracks. The high-magnification morphology shows that more fine quartz textures sized ~2–10 μm are evenly distributed on the hard semi-coke surface. These quartz particles are so tightly bound to the carbon matrix that it can be assumed that it is rather difficult for external forces to dislodge them from the carbon matrix. Additionally, owing to the high hardness of quartz, a large cohesive force needs to be overcome to crush it. It can be speculated that quartz in semi-coke acts as the skeleton of the semi-coke structure, which is the main reason for its low grindability and high abrasion.

Fig. 10.

Surface morphology of hard-to-grind semi-coke particles and distribution of quartz particles. (Online version in color.)

In the industrial experiment, it is worth noting that when 40% SA was added to CA, the theoretical and actual values of HGI of the mixed fuel were closer to each other, both of which are 67, which does not differ too much from the HGI value of 69 of PC, but has a significant impact on the grinding efficiency. The HGI test process involves spinning the fuel in the grinding bowl at a constant speed for 60 rpm, and the proportion of <71 μm is used for comparative evaluation. However, the growing trends of the proportion of <71 μm for semi-coke and PCI coal at higher cycles are different, as shown in Fig. 11. After the cycle exceeds 120, the proportions of the powders produced with the two semi-coke samples change slowly. However, as the number of cycle increases, the proportions of the powders produced with the two PCI coal samples still change rapidly. In an industrial mill, the residence time varies between different fuel types and particles of the same fuel, and as the residence time of some fuels increases, there may be a big difference between the grinding efficiency and predicted HGI for different fuels, which may be an important reason for the decrease in efficiency after the addition of semi-coke. This also implies that the current HGI test method may not be comprehensive enough to evaluate the grindability of semi-coke and may overestimate the grindability of semi-coke, and there is a need to explore a more reasonable predictor for grinding efficiency.

Fig. 11.

Variation of coal grinding size with grinding rotation number. (Online version in color.)

4.2. Method for Preparing Highly Grindable Semi-coke

The decrease in grinding efficiency and increase in process cost owing to the low grindability of semi-coke are the main factors limiting its use in blast furnace injection, and it is necessary to explore the feasibility of improving the grindability of semi-coke during its preparation. Since the ash constituents in raw coal will have a great impact on the HGI, full consideration should be given to the selection of coal types for pyrolysis that are suitable for blast furnace injection. We investigated 31 commercially available industrial semi-coke samples, analyzed the regression relationships between four main inorganic mineral constituents and HGI, and defined the mineral-impacted grindability index φ, which is expressed in Eq. (4)   

$$\phi =({\mathrm{m}}_{\mathrm{S}\mathrm{i}\mathrm{O}2}+{\mathrm{m}}_{\mathrm{A}\mathrm{l}2\mathrm{O}3})/({\mathrm{m}}_{\mathrm{C}\mathrm{a}\mathrm{O}}+{\mathrm{m}}_{\mathrm{F}\mathrm{e}2\mathrm{O}3})$$(4)

where m_SiO2, m_Al2O3, m_CaO, and m_Fe2O3 are the mass percentages of the corresponding oxides. Figure 12 shows the relationship between the mineral-impacted grindability index φ and HGI of raw coal. With increasing φ, the HGI gradually decreases, which is due to the fact that among the mineral constituents affecting grindability, siliceous minerals and aluminum minerals reduce the grindability (quartz has the most significant impact on it among siliceous minerals, followed by aluminum minerals), and calcium and iron minerals increase the grindability. The two semi-coke samples SA and SB herein also show the relationship between the two factors seen in Fig. 12. Based on this structural analysis, semi-coke with high HGI can be produced by preferentially selecting raw coal with lower φ.

Fig. 12.

Impacts of the mineral-impacted grindability index φ of semi-coke on HGI. (Online version in color.)

Additionally, the pyrolysis conditions have a significant impact on the HGI of semi-coke. Figure 13 shows the impact of the adjusted pyrolysis temperature on the HGI during the preparation of industrial semi-coke SA. As shown in Fig. 13(a), the HGI and pyrolysis temperature are negatively correlated, and increasing pyrolysis temperature is accompanied by a gradual decrease in HGI. Figure 13(b) shows the statistical results of pyrolysis temperature and HGI. It can be found that as the pyrolysis temperature increases, the HGI first increases, reaches the highest value near 640°C, and then decreases rapidly as the pyrolysis temperature increases. This may be due to the pyrolysis reaction that releases a large amount of volatiles, the development of the pore structure that forms cracks,³⁶⁾ and the decrease in matrix hardness. The thin walls of semi-coke pores and their low strength make it easy to remove them from the particle surface, which increases the grindability of semi-coke. As the pyrolysis temperature continues to increase, the semi-coke will undergo a pyrocondensation reaction, which will reduce its volume, shrink its pores, increase its particle density, and reduce its grindability. Therefore, for the same coal type, the highest value of HGI corresponds to a particular pyrolysis temperature. The SA in this industrial experiment corresponds to a high pyrolysis temperature, affording low HGI. The HGI can be further increased by lowering the pyrolysis temperature to ~640°C.

Fig. 13.

Grindability of industrial semi-coke during preparation: (a) pyrolysis furnace temperature and HGI monitored during different time periods; (b) pyrolysis furnace temperature vs. HGI. (Online version in color.)

In summary, the grindability of semi-coke can be improved via preferential selection of raw coal and control of pyrolysis conditions. We have used this idea to increase the HGI of industrial semi-coke to >62 to better meet the demand of a blast furnace for high grindability semi-coke, and will report the associated industrial experiments in the future.

4.3. Strategy for Adding Semi-coke into the Injection Fuel

As the grindability of existing industrial semi-coke is usually lower than that of conventional injection coal, its addition can reduce the output per machine-hour of the grinding workshop and increase electricity and energy consumption. Therefore, the amount of semi-coke should be controlled for the grinding workshop with a grinding capacity insufficient for the blast furnace. With low semi-coke addition, the impact on the grindability of the coal/semi-coke mixed fuel is in accordance with the weighted calculated value, and the use of semi-coke with highly grindable coals should be considered.

The grinding efficiency of the mill is closely related to the particle size of the product, which can be increased by appropriately increasing the semi-coke particle size. When the particle size is increased, the combustion performance of semi-coke SA in the raceway may decrease, but when the average particle size is <100 μm, its ignition temperature is still lower than that of pulverized coal CA (Fig. 14). Therefore, it is expected that appropriately increasing the semi-coke particle size will not adversely affect the combustion performance in the raceway. In industrial experiments, 40% semi-coke addition caused an overall increase in the particle size of the fuel injected into the blast furnace, but did not adversely affect blast furnace operation, suggesting that it is feasible to appropriately increase the particle size.

Fig. 14.

Impact of increased particle size on the ignition temperature of semi-coke. (Online version in color.)

From the perspective of heat release and combustion of the injection fuel in the raceway, the theoretical heat output of semi-coke itself is slightly lower than that of anthracite for injection, but its high combustion performance makes its actual combustion rate in the raceway of the blast furnace higher than that of PCI coal, thus increasing its effective heat output. This increased the coal/coke replacement ratio for the blast furnace in the industrial experiments, increased the coal ratio, reduced the unburned char, improved the permeability of the feed layer, and reduced the total fuel consumption, as evidenced by the lower carbon content in the blast furnace dust ash (Fig. 15). By contrast, the lower sulfur content in semi-coke reduced the amount of sulfur and fuel consumption. Therefore, injection of semi-coke with a high combustion reactivity index can help improve the furnace operation at different temperatures and reduce the fuel consumption and iron-making cost. Semi-coke has a bright application prospect in replacing the scarce anthracite coal currently used in blast furnaces.

Fig. 15.

Relationship between the ignition temperature and carbon content in blast furnace dust and coal injection ratio. (Online version in color.)

5. Conclusions

(1) The grindability of semi-coke is significantly lower than that of PCI coal. Semi-coke contains a high percentage of hard-to-grind quartz minerals that are extremely difficult to break during the grinding process, which can significantly reduce the grindability of coal and increase abrasion on the mill. When the proportion of semi-coke is low (≤40%), as it increases, the grindability and abrasion of semi-coke are consistent with the law of weighting. The addition of semi-coke will reduce the production of the mill, increase its differential pressure, and cause the grinding products to shift toward the larger-size region. Hard-to-grind semi-coke will reduce the grinding efficiency of the mill.

(2) The well-developed pore structure, large specific surface area, and surface morphologies of semi-coke can improve the kinetic diffusion of the combustion process. The lower aromatic layer stacking height and size of semi-coke indicate that it is slightly less ordered than PCI coal. The rich defective microcrystalline structure of semi-coke provides more reactive sites for combustion reactions.

(3) The addition of semi-coke to the injection fuel will decrease the fuel ratio of the blast furnace, increase the injection rate, improve the permeability of the blast furnace, reduce the carbon content in the dedusting ash, and improve the blast furnace operation. The high combustion performance of semi-coke makes its actual combustion rate in the raceway of the blast furnace higher than that of PCI coal, thus increasing its effective heat output. Appropriately increasing the semi-coke particle size can reduce the grinding cost without adversely affecting the combustion performance in the raceway.

(4) The lack of semi-coke grindability is a key factor limiting its large-scale use in blast furnace injection. As the difference between the pulverization yield of semi-coke and coal widens with increasing grinding time, the Hardgrove grindability index evaluation method is inadequate to evaluate the grindability of semi-coke. As the mineral-impacted grindability index φ of raw coal increases, the HGI of prepared semi-coke will decrease, which can be further increased by maintaining an appropriate pyrolysis temperature.

Acknowledgements

The authors gratefully acknowledged the National Natural Science Foundation of China (No. 51874224), Shaanxi Provincial Department of Education service local special project (22JC042) and Key Research and Development Program of Shannxi (No. 2021GY-128).

References

• 1)   S.  Han,  H.  Chen,  R. Y.  Long and  X. T.  Cui: Resour. Conserv. Recycl., 129 (2018), 293. https://doi.org/10.1016/j.resconrec.2016.08.012
• 2)   W. J.  Duan,  Q. B.  Yu,  H. Q.  Xie and  Q.  Qin: Energy, 135 (2017), 317. https://doi.org/10.1016/j.energy.2017.06.132
• 3)   C.  Zou,  J. Y.  He,  J. X.  Zhao,  X. M.  Li,  R. M.  Shi,  C.  Ma,  Y.  Kang and  X. R.  Zhang: Metall. Mater. Trans. B, 50 (2019), 2304. https://doi.org/10.1007/s11663-019-01647-4
• 4)   C. C.  Huang,  X. J.  Ning,  G. W.  Wang,  J. L.  Zhang,  Z. F.  Peng and  H. P.  Teng: J. Iron Steel Res. Int., 28 (2021), 391. https://doi.org/10.1007/s42243-020-00480-3
• 5)   S.  Gupta,  V.  Sahajwalla and  J.  Wood: Energy Fuels, 20 (2006), 2557. https://doi.org/10.1021/ef060271g
• 6)   C.  Wang,  P.  Mellin,  J.  Lövgren,  L.  Nilsson,  W. H.  Yang,  H.  Salman,  A.  Hultgren and  M.  Larsson: Energy Convers. Manag., 102 (2015), 217. https://doi.org/10.1016/j.enconman.2015.04.013
• 7)   J. L.  Zhang,  S.  Ren,  B.  Su,  Y. H.  Lin and  S. G.  Long: J. Iron Steel Res. Int., 19 (2012), 12. https://doi.org/10.1016/S1006-706X(12)60053-3
• 8)   A.  Babich,  D.  Senk,  M.  Knepper and  S.  Benkert: Ironmaking Steelmaking, 43 (2016), 11. https://doi.org/10.1179/1743281215Y.0000000042
• 9)   Y.  Serdar: Int. J. Coal Prep. Util., 41 (2019), 233. https://doi.org/10.1080/19392699.2019.1605989
• 10)   Q. C.  Lu,  W. N.  Xie,  F. B.  Zhang,  Y. Q.  He,  C. L.  Duan,  S.  Wang and  X. N.  Zhu: Fuel, 264 (2020), 116829. https://doi.org/10.1016/j.fuel.2019.116829
• 11)   H. B.  Vuthaluru,  R. J.  Brooke,  D. K.  Zhang and  H. M.  Yan: Fuel Process. Technol., 81 (2003), 67. https://doi.org/10.1016/S0378-3820(03)00044-4
• 12)   W. R.  Zuo,  Y. M.  Zhao,  Y. Q.  He,  F. N.  Shi and  C. L.  Duan: Int. J. Min. Sci. Technol., 22 (2012), 121. https://doi.org/10.1016/j.ijmst.2011.07.011
• 13)   A. S.  Trimble and  J. C.  Hower: Int. J. Coal Geol., 54 (2003), 253. https://doi.org/10.1016/S0166-5162(03)00039-9
• 14)   J.  Duan,  Q. C.  Lu,  Z. Y.  Zhao,  X.  Wang,  Y. X.  Zhang,  J.  Wang,  B.  Li,  W. N.  Xie,  X. L.  Sun and  X. N.  Zhu: Minerals, 10 (2020), 230. https://doi.org/10.3390/min10030230
• 15)   S. S.  Matin,  J. C.  Hower,  L.  Farahzadi and  S. C.  Chelgani: Int. J. Miner. Process., 155 (2016), 140. https://doi.org/10.1016/j.minpro.2016.08.015
• 16)   J. F.  Zhu,  J. Z.  Liu,  J. H.  Wu,  J.  Cheng,  Z. H.  Wang,  J. H.  Zhou and  K. F.  Cen: Fuel, 162 (2015), 305. https://doi.org/10.1016/j.fuel.2015.09.028
• 17)   D. V.  Miroshnichenko,  N. A.  Desna,  V. V.  Koval and  S. V.  Fatenko: Koks Khim., 62 (2019), 1. https://doi.org/10.3103/S1068364X19010058
• 18)   S.  Dindarloo,  J. C.  Hower,  A.  Bagherieh and  A. S.  Trimble: Int. J. Miner. Process., 154 (2016), 94. https://doi.org/10.1016/j.minpro.2016.07.005
• 19)   A. E.  Hansen and  J. C.  Hower: Int. J. Coal Geol., 131 (2014), 109. https://doi.org/10.1016/j.coal.2014.06.010
• 20)   S.  Hu,  L. Y.  Shan,  B. Y.  Xu,  J. L.  Su,  Y.  Wang,  L.  Jiang,  S.  Su and  J.  Xiang: Meitan Xuebao, 43 (2018), 855 (in Chinese). https://doi.org/10.13225/j.cnki.jccs.2017.0970
• 21)   A.  De Girolamo,  A.  Grufas,  I.  Lyamin,  I.  Nishio,  Y.  Ninomiya and  L.  Zhang: Energy Fuels, 30 (2016), 1858. https://doi.org/10.1021/acs.energyfuels.5b02425
• 22)   H. B.  Jiang,  J. L.  Zhang,  J. X.  Fu,  J.  Chang and  L.  Jing: J. Iron Steel Res. Int., 18 (2011), 6. https://doi.org/10.1016/s1006-706x(11)60029-0
• 23)   A. H.  Bagherieh,  J. C.  Hower,  A. R.  Bagherieh and  E.  Jorjani: Int. J. Coal Geol., 73 (2008), 130. https://doi.org/10.1016/j.coal.2007.04.002
• 24)   H. Y.  Li,  L.  Elliott,  H.  Rogers,  P.  Austin,  Y. L.  Jin and  T.  Wall: Energy Fuels, 26 (2012), 4690. https://doi.org/10.1021/ef201779q
• 25)   S. W.  Zheng,  Y. J.  Hu,  Z. Q.  Wang and  X. X.  Cheng: J. Energy Inst., 93 (2020), 1373. https://doi.org/10.1016/j.joei.2019.12.007
• 26)   C.  Moon,  Y.  Sung,  S.  Ahn,  T.  Kim,  G.  Choi and  D.  Kim: Appl. Therm. Eng., 54 (2013), 111. https://doi.org/10.1016/j.applthermaleng.2013.01.009
• 27)   J. M.  Steer,  R.  Marsh,  D.  Morgan and  M.  Greenslade: Fuel, 160 (2015), 413. https://doi.org/10.1016/j.fuel.2015.07.094
• 28)   S. W.  Du,  W. H.  Chen and  J. A.  Lucas: Energy, 35 (2010), 576. https://doi.org/10.1016/j.energy.2009.10.028
• 29)   H.  Lorenz,  E.  Carrea,  M.  Tamura and  J.  Haas: Fuel, 79 (2000), 1161. https://doi.org/10.1016/S0016-2361(99)00259-8
• 30)   Y.  Liu,  P. F.  Fu,  B.  Zhang,  F.  Yue,  H. C.  Zhou and  C. G.  Zheng: Fuel, 181 (2016), 1244. https://doi.org/10.1016/j.fuel.2016.01.077
• 31)   J.  Xu,  H.  Tang,  S.  Su,  J. W.  Liu,  K.  Xu,  K.  Qian,  Y.  Wang,  Y. B.  Zhou,  S.  Hu and  A. C.  Zhang: Appl. Energy, 212 (2018), 46. https://doi.org/10.1016/j.apenergy.2017.11.094
• 32)   C.  Zou,  J. X.  Zhao,  X. M.  Li and  R. M.  Shi: J. Therm. Anal. Calorim., 126 (2016), 1469. https://doi.org/10.1007/s10973-016-5806-y
• 33)   J. Y.  He,  C.  Zou,  J. X.  Zhao,  D.  Liang,  J. L.  Xi and  C.  Ma: J. Iron Steel Res. Int., 5 (2022), 727. https://doi.org/10.1007/s42243-021-00726-8
• 34)   A.  Ürünveren,  M.  Altıner,  O. B.  Ural and  S.  Ural: Energy Source., Part A, 39 (2017), 1216. https://doi.org/10.1080/15567036.2017.1315756
• 35)   C.  Ma,  C.  Zou,  J. X.  Zhao,  R. M.  Shi,  X. M.  Li,  J. Y.  He and  X. R.  Zhang: Energy Fuels, 33 (2019), 6098. https://doi.org/10.1021/acs.energyfuels.9b00860
• 36)   C.  Ma,  C.  Zou,  J. X.  Zhao,  J. Y.  He and  X. R.  Zhang: Thermochim. Acta, 688 (2020), 178576. https://doi.org/10.1016/j.tca.2020.178576