Deposit Formation in a Coal-Fired Rotary Kiln for Fluxed Iron Ore Pellet Production: Effect of MgO Content

Abstract

During the roasting process of fluxed pellets in a coal-fired rotary kiln, the incomplete combustion of pulverized coal injection accelerates deposit formation, which further limits the production efficiency of fluxed pellets. In order to eliminate the above problem, this study investigated the influence of MgO on deposit formation mechanism. The thermodynamic analysis revealed that MgO could increase the melting temperature of silicates in fluxed pellets with 0.8–1.2 basicity (CaO/SiO₂) when roasted at 1200–1250 °C, thereby decreasing the amount of liquid phase that formed initial deposits. XRD and SEM analyses of deposit simulants demonstrated that the addition of MgO was conducive to form magnesium magnetite and ferri-diopside, thereby avoiding the formation of hedenbergite with lower melting temperature. Moreover, the softening-melting performance and adhesivity tests confirmed that MgO had a positive effect on reducing liquid-phase deposition and inhibiting the adhesion of deposits on refractory bricks below 1250 °C. The above studies indicated that the addition of MgO helped to slow down the deposit formation of fluxed pellets prepared by coal-fired rotary kiln.

1. Introduction

In order to obtain high-quality ironmaking charges with better metallurgical performance and lower energy consumption, some pelletizing corporations have begun to expand the production scale of fluxed iron ore pellets in recent years to replace acid pellets and sinters [1,2]. For pelletizing corporations with sufficient coal supply capacity, they try to use coal-fired rotary kiln to produce fluxed pellets on a large scale. The rotary kiln can provide stable heat by the combustion of pulverized coal injection and roast pellets more uniformly by rotating around the central axis, which promotes the widespread application of coal-fired rotary kilns for pellet production [3,4]. However, deposit formation is considered to be a common obstacle in the production process of iron ore pellets by coal-fired rotary kiln [5]. The rapid accumulation of deposits on the kiln lining can seriously affect fluxed pellet production [6,7].

Until now, it has been revealed that the main factors leading to deposit formation include coal ash deposition, incomplete combustion of pulverized coal injection, roasting temperature fluctuations, low strength of preheated pellets, and high content of alkali metal in the raw materials [8,9]. The combustion products of pulverized coal injection in a coal-fired rotary kiln are mainly composed of coal ash and carbon residue [10]. Coal ash consists of aluminosilicate, which is one of the main initial reactants for deposit formation [11]. Some of the coal particles will not burn completely before they scatter to the rolling pellet layer, resulting in the presence of carbon residue in combustion products. These combustion products of coal are mixed with the powder produced by preheated pellet friction and collision, which continuously deposit below the pellet layer. The continuous combustion of residual carbon produces local reducing atmosphere in coal ash and pellet powder, which will promote the production of low-melting compounds containing Fe²⁺ [12,13,14]. These low melting point compounds composed of Si, Fe, Al, Ca, K, and Na generate liquid phase with good fluidity during high-temperature roasting [15,16], which adhere on the kiln lining and further erode the refractory bricks to form initial deposits [17,18]. Subsequently, the newly formed deposits continually adhere coal ash and pellet powder from the pellet layer. As the running time of the rotary kiln accumulates, the deposits are continuously thickened according to the “layer-by-layer” formation theory, ultimately dividing into loose layer adhered pellets and tight layer with a high proportion of silicate glassy phase [19]. The formation mechanisms of deposits are microcrystalline junction of hematite and liquid-phase binding [20]. Notably, deposit formation is inseparable from the promotion of FeO produced by residual carbon reduction. The silicate glassy phase generated by the reaction between FeO and silicate is the main component of the newly formed deposits. Owing to the better adhesive performance of the silicate glassy phase, the initial deposits arrange closely and are difficult to be destroyed [21]. With the thickening of deposits, more preheated pellet powder adheres on them. Depending on the recrystallization of hematite, the strength of deposits further increases [22]. In addition, the fluctuation of roasting temperature is also one of the main inducements of deposit formation [23,24,25]. Dense and thick deposits usually occur in the higher temperature area of rotary kiln [26].

In summary, the above previous studies are only aimed at the deposit-forming mechanism of acid pellets. Although some pelletizing corporations produce fluxed pellets by grate-kiln process at present, there is also a lack of corresponding research on the deposit formation mechanism of fluxed pellets. The CaO in fluxed pellets reacts with Fe₂O₃ to form calcium ferrite phase, which melts into liquid phase to promote the consolidation of hematite particles during the roasting process [27,28]. Therefore, preheated fluxed pellets themselves form liquid phase during the roasting process to accelerate the deposit formation [29]. To reverse this unfavorable situation, we added MgO powder into the raw material used to prepare fluxed pellets. The addition of magnesium usually increases the melting temperature of low-melting-point compounds [30]. Based on this theory, MgO powder can be used to slow down the deposit formation. In addition, with regards to fluxed pellets, adding a suitable amount of MgO is also beneficial to improve their metallurgical properties [31]. However, the MgO content in the raw materials should not exceed 3 wt% to guarantee the strength of fluxed pellets [32].

In order to eliminate the harmful effects of deposits on the production of fluxed pellets by coal-fired rotary kilns, it was necessary to further study the influence of MgO on deposit formation. In this paper, Factsage software was used to calculate the substance composition and liquid-phase content of fluxed pellets with different MgO additions at 1100–1300 °C. Furthermore, preheated pellet powder, removed volatile pulverized coal (RVC), and self-prepared coal ash (SCA) were used to prepare deposit simulants. The phase composition and microstructure of deposit simulants were analyzed by XRD and SEM-EDS. Subsequently, the adhesivity of deposit simulants to kiln lining refractory materials were presented, and the softening-melting performance of deposit simulants at 1200–1300 °C were investigated. The above detailed experiments were performed to verify the inhibiting effect of MgO on deposit formation during fluxed pellet production by coal-fired rotary kilns. Meanwhile, these research findings provided theoretical and technical guidance for efficient production of fluxed pellets by grate-kiln process.

2. Materials and Methods

2.1. Materials

The following raw materials was obtained from an ironmaking plant that used a coal- fired rotary kiln technological process to produce fluxed pellets.

Table 1. Chemical compositions of raw materials (wt%).

Raw Material	TFe	FeO	SiO2	CaO	Al2O3	MgO	TiO2	K2O	Na2O	S
iron concentrate	64.83	9.82	3.55	0.50	1.07	0.77	-	0.06	0.06	0.04
Preheated pellet	62.10	-	4.35	4.20	1.32	0.81	-	0.09	0.08	0.04
Limestone	0.52	0.14	2.32	51.82	0.46	2.63	-	0.17	0.03	0.03
Magnesium oxide	0.01	-	0.07	0.01	-	99.60	-	-	-	0.19
Bentonite	3.09	0.20	60.01	3.73	12.52	3.00	-	1.40	1.80	0.02
Deposit sample	59.74	0.76	6.67	2.98	2.93	0.69	-	0.24	0.17	0.02
Refractory bricks	1.23	-	16.98	1.75	71.26	0.24	1.89	-	-	-
Coal ash	4.04	0.10	51.85	8.57	26.09	1.12	-	1.22	0.75	0.58

shows the chemical compositions of iron concentrate, preheated pellet powder, limestone, pure magnesium oxide, bentonite, deposit sample, refractory bricks, and coal ash, respectively. In order to study the effect of incomplete combustion pulverized coal on deposit formation, removed volatile pulverized coal (RVC) and self-prepared coal ash (SCA) were used to simulate the composition of incomplete combustion pulverized coal. The proximate analysis results of pulverized coal, removed volatile pulverized coal (RVC) and self-prepared coal ash (SCA) are presented in

Table 2. Proximate analysis results of pulverized coal, RVC, and SCA (%).

Raw Material	Mad, %	Aad, %	Vad, %	FCad, %
Pulverized coal	1.81	12.89	24.34	60.96
Removed volatile pulverized coal (RVC)	0.70	16.52	1.94	80.84
Self-prepared coal ash (SCA)	0.14	99.44	0.41	0.01

. Since the volatiles were preferentially removed at high temperature, RVC with 80.84% fixed carbon could be prepared by roasting pulverized coal in a closed corundum crucible at 920 °C for one hour (N₂ atmosphere). Furthermore, SCA was obtained by roasting pulverized coal at 815 °C for 4 h (air atmosphere).

2.2. Methods

2.2.1. Preparation of the Initial Reactants for Deposit Formation

Earlier studies have confirmed that the initial deposits of coal-fired rotary kiln were the roasting products of preheated pellet powder and incomplete burnt pulverized coal [33]. In order to further study the deposit formation mechanism, we needed to synthesize the above initial reactants (preheated pellet powder and incomplete burnt pulverized coal). The first step was to prepare preheated pellet powder. According to the prescribed binary basicity (CaO/SiO₂) and magnesium content, the iron concentrate was mixed with limestone, pure magnesium oxide, and 1.2 wt% bentonite in proportion. The raw materials were pelletized into 12–14 mm green pellets. Then a tube furnace was used to preheat these green pellets for 6 min at 1070 °C. After preheating, the pellets were grinded in a rotary drum machine to collect surface powder for subsequent works. The next step was to prepare incomplete burnt pulverized coal. Our previous research summarized the influence of pulverized coal with different burn-off rates on deposit formation, and expounded the synthetic method for incomplete burnt pulverized coal [34,35]. Herein, pulverized coal with 90% burn-off rate was selected for the following experiments, because it was similar to the actual burning condition of pulverized coal. Due to the incomplete burnt pulverized coal being composed of coal ash and fixed carbon, we mixed SCA with RVC to simulate the incomplete burnt pulverized coal according to Equations (1)–(3). Furthermore, the composition of incomplete burnt pulverized coal with 90% burn-off rate mixed by SCA and RVC are shown in

Table 3. Composition of simulated pulverized coal with 90% burn-off rate.

Burn-Off Rate/%	SCA/wt%	RVC/wt%
90	60.61	39.39

.

(1)

Burn-off rate of pulverized coal, which was calculated from Equation (1):

$$\mathrm{Burn} \mathrm{off} \mathrm{rate} \left(\mathsf{\eta }\right)=\frac{\mathrm{m} (\mathrm{Combustion} \mathrm{part} \mathrm{pulverized} \mathrm{coal}) }{\mathrm{m} (\mathrm{The} \mathrm{amount} \mathrm{of} \mathrm{pulverized} \mathrm{coal} \mathrm{input})}\times 100\%$$

(1)

(2)

SCA percentage of pulverized coal input, which was worked out by Equation (2):

$$\mathrm{SCA} \mathrm{percentage}=\frac{\mathsf{\eta }\times 12.89\%}{99.44\%}$$

(2)

(3)

RVC percentage of pulverized coal input, which was calculated by Equation (3):

$$\mathrm{RVC} \mathrm{percentage}=\frac{(100\%-\mathsf{\eta })\times 60.96\%}{80.84\%}$$

(3)

The deposit simulants were obtained by roasting the initial reactants of deposits at high temperature. The deposit simulants with the same element content as the deposit sample in Table 1 could be obtained by calculation. The equilibriums of chemical elements between pellet powder, coal ash, and deposit samples were determined by the following Equations (4) and (5). Calculation results indicated that 9.434 g preheated pellet powder and 0.566 g coal ash were needed to prepare 10 g deposit simulants. Based on the mass percentage in Table 3 and Equation (6), the respective weights of SCA and RVC used to prepare 10 g deposit simulants are listed in

Table 4. The respective weights of the initial reactants for simulating deposit formation.

Preheated Pellet Powder/g	SCA/g	RVC/g
9.434	0.334	0.514

.

$$\begin{array}{l}\mathrm{w} (\mathrm{TFe}/\mathrm{Si}/\mathrm{Al}) \mathrm{pellet} \mathrm{powder}\times \mathsf{\alpha } (\mathrm{pellet} \mathrm{powder})+\mathrm{w} (\mathrm{TFe}/\mathrm{Si}/\mathrm{Al}) \mathrm{coal} \mathrm{ash}\times \mathsf{\beta } (\mathrm{coal} \mathrm{ash})\\ =\mathrm{w} (\mathrm{TFe}/\mathrm{Si}/\mathrm{Al}) \mathrm{deposit}\end{array}$$

(4)

$$\mathsf{\alpha } (\mathrm{pellet} \mathrm{powder})+\mathsf{\beta } (\mathrm{coal} \mathrm{ash})=100\%$$

(5)

$$\mathsf{\beta } (\mathrm{coal} \mathrm{ash})=\mathrm{w} (\mathrm{SCA})\times 99.44\%+\mathrm{w} (\mathrm{RVC})\times 16.52\%$$

(6)

where w (TFe/Si/Al) pellet powder, w (TFe/Si/Al) coal ash, w (TFe/Si/Al) rotary kiln deposit denoted the contents of Fe, Si, and Al in preheated pellet powder, coal ash, and deposit samples, respectively; α (pellet powder) and β (coal ash) were the mass percentage of preheated pellet powder and coal ash, respectively. W (SCA) and w (RVC) represented the mass percentage of removed volatile pulverized coal (RVC) and self-prepared coal ash (SCA) in pulverized coal with 90% burn-off rate.

2.2.2. Preparation of Deposit Simulants

To study the effect of MgO on the phase composition and microstructure of deposits, we needed to prepare deposit simulants with different MgO content. The initial reactants of deposit simulants mixed by preheated pellet powder and incomplete burnt pulverized coal with 90% burn-off rate were pressed into cylindrical clumps at 10 mm diameter. Then, the dried cylindrical clumps were placed into a muffle furnace and roasted in air atmosphere at 1250 °C for 2 h. After cooling, the deposit simulants for X-ray diffractometer and scanning electron microscope analysis were obtained.

2.2.3. Evaluation Method for Softening-Melting Performance

The evaluation on softening-melting performance of deposit simulants was used to characterize the difficulty of deposit formation. The initial reactants (preheated pellet powder and incomplete burnt pulverized coal) of deposits containing fixed carbon would produce liquid phase with better fluidity during roasting. It could be judged that the less solid residue left after the initial reactants melting, the more liquid phase would be generated to form initial deposits eventually. Consider these above reasons, ash fusion apparatus was used to record the area residue of the pyrometric cones during the roasting process. This method could characterize the generation of molten liquid phase from the initial reactants of deposit simulants according to the standard ISO 540:2008(E) and ASTM D1857/D1857M-16. A schematic diagram of ash fusion apparatus is shown in Figure 1. Firstly, we prepared pyrometric cones by the initial reactants of deposit simulants with different MgO contents. Then, the pyrometric cones were placed into the thermocouple temperature measuring area of ash fusion apparatus. The temperature measuring area was heated from room temperature to 1300 °C with 5 °C/min heating rate. photographs of the pyrometric cones were taken every 10 s starting from 900 °C. As the temperature increased sequentially, the molten liquid phase of initial reactants flowed downward under the influence of gravity; thus, the residual area of pyrometric cones deceased gradually. After roasting, we selected the photographs of pyrometric cones taken at 1000 °C, 1200 °C, 1250 °C, and 1300 °C, respectively. Then, Photoshop CS6 software was used to calculate the pixel numbers of each pyrometric cone in the photographs. It was worth noting that each pixel in the photos represented a micro-region with the same area. The sum of pixels could be used to replace the residual area of pyrometric cones. Moreover, the area residual rate of the pyrometric cones at different temperatures could be calculated according to Equation (7) [36]:

$$\mathsf{\delta } (1200/1250/1300 °C) \mathrm{area} \mathrm{residual} \mathrm{rate}=\frac{\mathrm{n} (1000 °C) \mathrm{pixels}-\mathrm{n} (1200/1250/1300 °C) \mathrm{pixels}}{\mathrm{n} (1000 °C) \mathrm{pixels}}$$

(7)

where δ (1200/1250/1300 °C) area residual rate was the residual area percentage of pyrometric cones at 1200 °C, 1250 °C, and 1300 °C, respectively, compared to the cone area at 1000 °C; n (1000 °C) pixels was the pixels number of each pyrometric cone in the photographs at 1000 °C; and n (1200/1250/1300 °C) pixels was the pixels number of each pyrometric cone in the photographs at 1200 °C, 1250 °C, and 1300 °C, respectively.

2.2.4. Evaluation Method for Adhesivity to Refractory Brick

When the newly formed deposits exhibited adhesivity to the refractory brick during roasting, the growth of rotary kiln deposits would be accelerated. Hence, it was necessary to investigate the influence of MgO on the adhesivity of deposits to refractory bricks. We selected the refractory bricks of rotary kiln lining and measured their weight, respectively. After that, the 90% burn-off rate initial reactants (preheated pellet powder and incomplete burnt pulverized coal) of deposit simulants (3 g) with different MgO mass fractions were laid on the upper surface of refractory bricks, and their contact surfaces were ensured to be a circle with the same area. The above samples (initial reactants and refractory bricks) were roasted in a muffle furnace at 1250 °C for 5 h to obtain refractory-brick-adhered deposit simulants. Generally, during the roasting process of the rotary kiln, the deposits would be squeezed and rubbed by rolling pellets and could not completely attach on the kiln lining. Therefore, as shown in Figure 2, we selected a 3 kg roller to drag on the surface of samples in the horizontal direction of refractory bricks three times, with a constant speed (10 cm/s), to simulate the collision and rolling process. After the loose deposit pieces were removed, the residual deposits firmly bonded to the refractory bricks. Meanwhile, the current weight (deposit simulants and refractory bricks) of the refractory bricks was measured to calculate the residual weight of deposit simulants [35].

2.3. Analysis and Characterization

The proximate analysis of pulverized coal, removed volatile pulverized coal (RVC), and self-prepared coal ash (SCA) were tested according to a Chinese standard GB/T 212–2008. The preparation and detection of pyrometric cones were based on national standards ISO 540:2008(E) and ASTM D1857/D1857M-16. The phase composition of the deposit simulant with different MgO content was measured by X-ray diffractometer (XRD, Empyrean, Malvern PANalytical, Egham, Britain). The database of XRD analysis used PDF2-2004 version. The microstructure of the deposit simulant with different MgO content was characterized using a field emission scanning electron microscope (SEM, TESCAN MIRA3, TESCAN, Brno, The Czech Republic) coupled with an energy dispersive spectra analyzer (EDS, Oxford X-max 20, Oxford Instruments, Oxford, Britain). Thermodynamic software FactSage 8.0 was used to analyze the effect of MgO content on fluxed pellet composition and liquid-phase content. Although the Factsge software has the problem of missing database in some special fields, the results of the phase diagram calculation of fluxed pellets are reliable. In this software, we selected corresponding databases “FToxid” and “FactPS”. Moreover, “Phase diagram” function was used to draw ternary phase diagrams at different calcination temperatures, and “Equilib” function was used to simulate the change in liquid phase amount and other component contents of fluxed pellets based on the raw material composition [37].

3. Results

3.1. Phase Diagram Calculation of Preheating Fluxed Pellet Powder

The oxidizing atmosphere was commonly used in rotary kilns to prepare fluxed pellets. In addition to incomplete combustion of pulverized coal resulting in the formation of liquid phase, the fluxed preheated pellet powder itself would still produce liquid phase to accelerate the formation of initial deposits during high temperature roasting. We calculated the phase diagrams of Fe₂O₃–CaO–SiO₂ system with different MgO content by “Phase diagram” function of Factsage8.0 software from 1100 °C to 1300 °C. As revealed by Figure 3a, Fe₂O₃–CaO–SiO₂ phase diagram exhibited that calcium iron compounds, including Ca₃Fe₂Si₃O₁₂, CaFe₂O₄, and CaFe₄O₇, had low melting temperature. For fluxed pellets, these low-melting-point compounds would induce deposit formation. Therefore, increasing the melting temperature of these compounds was an effective method to restrain the deposit formation. We continued to study the effect of MgO addition on the composition of calcium iron compounds. Figure 3b shows a Fe₂O₃–CaO–SiO₂–MgO quaternary system phase diagram with 2 wt% MgO. From this figure, we could see that Mg first appeared in silicates (Ca₃MgSi₂O₈). As the MgO content reached 3 wt%, more magnesium-containing silicates appeared in Figure 3c, including CaMgSi₂O₆ and Ca₂MgSi₂O₇, which indicated that MgO reacted easily with silicates in the temperature range of 1100–1300 °C. These calculation results revealed that the addition of MgO could directly affect the substance composition of deposits.

“Equilib” function of Factsage8.0 software was used to calculate the amount of liquid phase in fluxed pellets with different MgO content at 1100–1300 °C. According to the actual mass fraction of Fe₂O₃, CaO, SiO₂, and Al₂O₃ in preheated pellet (88.71 wt% Fe₂O₃, 4.47 wt% (CaO + SiO₂), and 1.32 wt% Al₂O₃), the mass fraction of MgO was set as 0 wt%, 2 wt%, and 3 wt%, respectively. As shown in Figure 4a, when roasting temperature reached above 1200 °C, andradite in 0.8 basicity pellets could be melted into liquid phase. The amount of liquid phase increased gradually with the roasting temperature from 1200 °C to 1300 °C. In Figure 4b,c, the addition of MgO promoted the formation of magnesium compounds, including spinel, clinopyroxene, and melilite, and inhibited the formation of liquid phase at 1200–1250 °C. Only when the roasting temperature reached 1300 °C, clinopyroxene and melilite would be converted into the liquid phase. As revealed by Figure 4d, rankinite and 2.92 wt% liquid phase would generate during the roasting process of fluxed pellets with 1.0 basicity at 1250°C. With the addition of MgO, the liquid phase disappeared below 1250 °C. Meanwhile, more melilite and spinel would generate. Neither of them would melt below 1250 °C (Figure 4e,f). In Figure 4g, when the basicity rose to 1.2, the liquid phase amounts of fluxed pellets without MgO addition increased to 8.01 wt% at 1250 °C. When the MgO content reached 2–3 wt%, the formation of liquid phase would be restrained (Figure 4h,i). In summary, when the preheated pellet powder with 0.8–1.2 basicity was roasted at 1200–1250 °C, this would produce liquid phase to accelerate the formation of rotary kiln deposits. Noteworthily, 2–3 wt% MgO was attributed to the generation of higher melting temperature magnesium compounds, including spinel and melilite, which would inhibit the formation of liquid phases.

3.2. XRD Analysis of Deposit Simulants

MgO could not only decrease the amount of liquid phase produced by fluxed pellets themselves during the roasting process, but also affect the deposit formation caused by incomplete combustion of pulverized coal. We further investigated the influence of MgO on rotary kiln deposits. X-ray diffraction technology was used to analyze the phase composition of deposit simulants prepared by preheated pellet powder with 1.0 binary basicity and incomplete combustion pulverized coal with 90% burn-off rate. As shown in Figure 5, hedenbergite was obtained by roasting the initial reactants of deposit simulants without MgO addition at 1250 °C for 2 h. Because of low melting point (approximately 1150 °C), hedenbergite was identified as one of the main components in fluxed pellets that would accelerate deposit formation. When the content of MgO increased to 2–3 wt%, ferri-diopside (the isomorphism of hedenbergite and diopside) and magnesium magnetite appeared in the deposit simulants. Since the Mg²⁺ ionic radius (0.078 nm) was close to Fe²⁺ (0.082 nm), these two ions could be replaced with each other in crystals and their oxides could form solid solution in any proportion. MgO reacted with Fe₃O₄ generated in reducing atmosphere to form solid solution or enter the silicates. By comparison, ferri-diopside and magnesium magnetite exhibited higher melting temperature than hedenbergite, which could inhibit the increase in liquid phase amounts during the roasting process. Therefore, increasing the MgO content was beneficial to solve the deposit forming problem of fluxed pellets caused by incomplete combustion of pulverized coal injection.

3.3. SEM-EDS Analysis of Deposit Simulants

We used SEM-EDS technology to further analyze the phase composition and microstructure of the deposit simulants with 2 wt% MgO. As revealed by Figure 6a, after roasting at 1250 °C for 2 h, the hematite particles were bonded together by silicates to form a microstructure with certain consolidation strength in magnesium-containing deposit simulants, which was similar to the deposits of calcium-fluxed pellets. However, the difference between fluxed pellets and magnesium-fluxed pellets was that the MgO addition caused the conversion from hedenbergite to ferri-diopside in deposit simulants. Figure 6b shows an SEM image of hematite (point 1) and magnesium magnetite (point 2). According to the atomic number proportion detected by EDS technology, it could be confirmed that magnesium magnetite (point 2) formed by the reaction between MgO and magnetite. Thereby, the addition of MgO would restrict part of Fe²⁺ into silicates, which decreased the production of low-melting-point hedenbergite. The SEM images of silicate binding phase in the deposit simulants are shown in Figure 6c. EDS analysis on point 3 exhibited that the Fe/Mg weight ratio reached 3.1, which demonstrated that one of the silicate binding phases in the deposit simulants with 2 wt% MgO was ferri-diopside. Moreover, there was a part of hedenbergite (point 4) which has not been converted into ferri-diopside in silicate bonding phase. It was also confirmed that the addition of MgO could change the chemical composition of silicate bonding phase and further increase the melting temperature of the low-melting-point compound that formed initial deposits.

3.4. Effect of MgO on Softening-Melting Performance of Deposits

The molten liquid phase formed by roasting the mixtures of preheated pellet powder and incomplete burnt pulverized coal would accelerate the deposit formation. Therefore, we tested the softening-melting performance of pyrometric cones prepared by preheated pellet powder and incomplete burnt pulverized coal at different roasting temperatures to characterize the generation of molten liquid phase. MgO contents refer to the mass fraction of MgO in the initial reactants of deposit simulants. As shown in Figure 7, the morphology of the pyrometric cones was photographed at 1000 °C, 1200 °C, 1250 °C, and 1300 °C, respectively. With the roasting temperature increasing from room temperature to 1300 °C, the combustion of residual carbon in pyrometric cones resulted in the formation of low-melting-point compounds. These compounds melted and gradually deposited downwards; meanwhile, the area of pyrometric cones shrank continuously. At 1250 °C, the area residual rate of pyrometric cones without extra MgO addition was 79.85%, while the area residual rate of cones containing 3 wt% MgO could reach 86.93%. It indicated that MgO had a positive effect to inhibit the formation of liquid phase and deposits at 1250 °C. Note that, when the roasting temperature increased to 1300 °C, the area residual rate of pyrometric cones containing 2–3 wt% MgO declined rapidly, which indicated ferri-diopside would be melted into liquid phase during roasting at 1250–1300 °C.

3.5. Effect of MgO on Adhesivity of Deposits to Refractory Brick

The deposits formed by roasting the mixtures of preheated pellet powder and incomplete burnt pulverized coal would adhere on the refractory materials and erode the surface of these materials. Generally, the stronger the interaction between deposits and refractory materials, the firmer the deposits would adhere on kiln lining. To study the effect of MgO on deposit adhesivity to refractory brick, 3 g initial reactants (preheated pellet powder and incomplete burnt pulverized coal) of deposits were placed on refractory brick of rotary kiln lining. Then, both initial reactants and refractory bricks were roasted for 5 h at 1250 °C. The weight of deposit simulants adhered to the refractory brick is shown in Figure 8. As the MgO content increased in initial reactants, the amounts of deposit simulants adhering on the surface of refractory brick decreased gradually after roasting. The adhesion weight of deposit simulants without MgO addition reached 0.13 g; by contrast, the initial reactants containing 3 wt% MgO only formed 0.07 g deposits adhered on refractory brick. To summarize, the liquid phase generated by residual carbon combustion powder promoted the adhesion of deposit simulants on the surface of refractory bricks. However, the addition of MgO could decrease the adhesion weight of deposits on the refractory brick, thus slowing down the accumulation rate of deposits.

4. Discussion

According to the results of thermodynamic software calculation, XRD and SEM-EDS analyses, the inhibition mechanism of MgO on deposit formation of fluxed pellets could be summarized. Figure 9 shows the schematic diagram of coal-fired rotary kiln and the deposit formation process of magnesium-containing fluxed pellets. During the roasting process of preheated pellets in the rotary kiln, the pellets collided and rubbed with each other to strip off surface powder containing hematite, andradite, and rankinite. The preheated pellet powder was mixed with the incomplete burnt pulverized coal containing fixed carbon as the initial reactants to form the deposits. Then, the thermal radiation in the rotary kiln made the fixed carbon burn to form local reductive atmosphere in initial reactants (preheated pellet powder and incomplete burnt pulverized coal) of deposits. Fe²⁺ produced by reduction reaction in initial reactants was absorbed by magnesium-containing compounds to form magnesium magnetite and ferri-diopside, which increased the formation temperature of liquid phase and weakened the adhesion of deposits to the kiln lining. Thus, it could be concluded that the MgO addition had a positive effect on slowing down the deposit formation in coal-fired rotary kiln for fluxed pellet production.

5. Conclusions

Herein, the inhibition of MgO on the deposit formation of coal-fired rotary kiln for fluxed pellet production was studied. MgO could not only reduce the amount of liquid phase produced by fluxed pellets themselves during the roasting process, but also inhibit the deposit caused by incomplete combustion of pulverized coal. The specific conclusions from the investigations in this paper were summarized as follows.

(1)

Based on the calculation results of Factsage8.0 software, fluxed pellet powder with 0.8–1.2 binary basicity roasted at 1200–1250 °C would produce 2.30–8.01 wt% liquid phase, which led to the formation of initial deposits. As MgO content increased to 2–3 wt%, the formation of these liquid phases in fluxed pellet powder could be inhibited at 1200–1250 °C.

(2)

The deposit simulants prepared by roasting the mixtures containing preheated pellet powder and incomplete burnt pulverized coal were mainly composed of hematite, magnesium magnetite, and ferri-diopside. The SEM-EDS analysis showed that the hematite and magnesium magnetite were bonded together by ferri-diopside to form the microstructure of deposit simulants. The Fe²⁺ formed by the combustion of residual carbon could be absorbed by magnesium oxide, which decreased the formation of low-melting-point hedenbergite, thereby inhibiting the decline in liquid phase formation temperature and slowing down the deposit formation.

(3)

With MgO added, the area residual rate of temperature cones prepared by preheated pellet powder and incomplete burnt pulverized coal at 1200–1250 °C was relatively higher than cones without extra MgO addition, which confirmed that MgO had a certain inhibitory effect on the formation of liquid phase. The results of the adhesivity experiment showed that the addition of MgO could reduce the adhesion weight of deposit simulants on the refractory bricks.