Effect of Carbonaceous Residues on the Blast Furnace Operation Efficiency by PC Injection

Efficiency of pulverized coal (PC) injection determines the coke rate and, consequently, the environmental effect, and depends significantly on its conversion. At high injection rates over 200 kg tHM−1, a significant part of coal residues, named char, leave the raceway. Char properties and behavior are of great importance in view of their possible consumption outside the raceway. This study continues the investigations of previous works reported in 2020, and comprises numerous laboratory, analytical tests and simulations performed at RWTH Aachen University, as well as industrial measurements. First, the article focuses on the char formation and the temperature development in the raceway while injecting PC. In addition, the effect of alkali on the char reactivity is proven. Next, the effect of injected PC on the pressure drop and permeability in different blast furnace (BF) zones is examined. Also, the char effect on liquid products and the dripping zone is studied. Hot‐metal carburization, dynamic and static holdups in the dripping zone, and slag characteristics are investigated. Further available carbonaceous residues in the BF, such as coke fines and soot, are investigated to reveal their competitive consumption. Finally, a method for identification and quantification of carbon phases in BF dust is developed.


Introduction
Counter current flow reactors are the most efficient metallurgical vessels. This is why the blast furnace (BF) remains since more than 700 years No. 1 aggregate for primary metal production. It is anticipated that also in the future, shaft furnaces operating on this principle will dominate the direct reduction (DR) production. Consequently, the gas permeability is a crucial issue for shaft furnaces. In the BF, required gas permeability, as well as a drainage ability in the dripping zone, is ensured, primarily, by a certain amount and quality of coke.
The injection of solid auxiliary reducing agents (ARA) may negatively affect the gas permeability and the drainage ability by two means: decrease of thickness of coke layers due to lower coke rates─and accumulation of unburnt injected particles outside the raceway─in dripping zone (DZ), cohesive zone (CZ), and shaft. Concerning the first reason, the burden-coke layer thickness ratio was increased, for example, from 0.93 for only coke operation to 1.54 while injecting 200 kg t HM À1 pulverized coal (PC) according to the calculations conducted. [1] Effects of layer thickness on the pressure drop in the BF "dry" zone and in the cohesive zone become apparent at increasing Reynolds number and flow rate. [2] Concerning the second reason, the mathematical modeling showed that char accumulates mainly in cohesive zone, deadman, and hearth. [3] The effect of accumulation of unburnt PC on the specific pressure drop is examined experimentally in this work.
Factors affecting the BF operation efficiency while injecting PC can be divided into four major groups: 1) characteristics and conversion behavior of coal and formed char; 2) characteristics of charged materials and liquid products, and their change caused by PC injection; 3) BF operation parameters; and 4) design of tuyere and injection lance.
These factors are also considered vital important by the current trend of PC co-injection with hydrogen-containing gases such as coke oven gas [4,5] or pure hydrogen. [6] Chemical, physical, mechanical, and petrographic characteristics of injected coals and their effect on the conversion behavior in the raceway, reducing gas, and heat generation and, consequently, on the coke replacement ratio are well known and summarized in Ref. [7,8] This also applies to the effect of BF operation parameters such as blast temperature, oxygen enrichment, top gas parameters, etc., as well as tuyere and lance arrangement. [7][8][9] The effect of PC on raceway adiabatic flame temperature (RAFT) is shown in Table 1 for coals with different calorific values.
The key problem, related to the PC injection efficiency, is that at high injection rates, unburnt coal particles leave the raceway despite numerous measures for intensifying their conversion, [7,8] including novel ones, such as introduction of "hot" and pulsed oxygen. [10][11][12][13] The question is whether the PC injection rate should not exceed the value which can be converted in the raceway or can we expect that char will be consumed outside the raceway, and if so, to which extent. It is also the question to which extent the values of RAFT changes presented in Table 1 are applicable in the case of incomplete PC conversion. It should be stressed that char is not char: it could be partly or completely devolatilized particles, or pyrolyzed ones, or already partly burnt coal particles. Char evolution and its kinds are described in Ref. [14].
This contribution is dealing with char and focuses on two first groups of aforementioned parameters.
Char formation in the raceway was already reported in Ref. [14] and showed that coal particles react predominantly from the inside and become a more porous structure; a char "skeleton" remains after coal devolatilization and initial conversion. Here, further examinations of char properties and alkalis effect on char reactivity have been undertaken, and PC effect, particularly the role of unburnt coal, on the temperature in the raceway has been investigated.
Char transport and behavior outside the raceway may occur mostly by following options: 1) interaction with and impact on coke and iron burden, 2) reaction with liquid slag, 3) dissolution in hot metal and its carburization, 4) reaction with CO 2 in shaft, 5) accumulation in different BF zones, and 6) transport through the furnace and identification in the BF dust.
Char effect on coke and iron burden has been studied in Refs. [3,14]. Most of further mentioned options are discussed in this contribution. In this context, the original study on the char effect on liquid products and the dripping zone should be highlighted.
It should also be taken into account that the coal char is not only a carbon fine in the BF. Coke fines (CF) and soot generated from PC or gaseous ARA like natural gas or coke oven gas are further carbon sources in the BF. Therefore, this contribution focuses also on the competitive consumption of these materials in-and outside the raceway, that is to examine which of the available carbon fines will react preferably.
Finally, the char remaining in the flue dust is the last witness of what happened with PC during its trip via the BF. Therefore, its identification and quantification are crucial for the control of the PC injection efficiency.

Facilities and Methods
A number of laboratory and analytical facilities and methods at RWTH Aachen University were used in this work.
For the examination of coal and CF conversion, and char generation (Section 3.1, 3.2), Multifunctional Injection Rig for Ironmaking (MIRI) and Tammann furnace drop tube setup (TF-DT) were used (Figures 1 and 2). These facilities are described in Refs. [3,9]. The MIRI plant simulates the conversion behavior of solid fines under the raceway conditions and TF-DT─under the predefined conditions, such as a fixed temperature. The residence time of PC within high-temperature zone of the reaction furnace makes up 20-30 ms depending on gas flow rate (similar to the raceway oxygen zone). [9] Conversion degree η of ARA in the MIRI trail can be calculated based on 1) off-gas analysis using the nitrogen balance, 2) ash content of original material and residue, and 3) carbon and ash contents Table 1. Decrease of RAFT while injecting PC. [7,8] Coal type Decrease in RAFT Anthracite 80°-90°C/100 kg t HM

À1
Subbituminous/lignite 180°-210°C/100 kg t HM Another Tammann furnace setup, shown in Figure 3 and described in Ref. [2], was used to determine the activation energy of coals, chars, CF, and soot (Section 3.1, 3.2) and to examine the alkali effect on char and coke reactivity (Section 3.2). The total gas flow rate is kept constant (200 L h À1 ) in all tests. The specimen is placed in a crucible (7) and suspended from an arm of the electronic balance with a platinum wire. The weight loss measured by electronic balance (3) is recorded continuously by software on an attached computer (4).
For temperature measurements in the raceway at industrial BFs (Section 3.3), a thermal vision camera (TVC) IVS 6400 from Impac Infrared, which operates in the near-infrared range at approximately 1 mm, was applied. An easily moveable platform was used to ensure measurements at different tuyeres and BFs. The system ensures the measuring frequency of 60 Hz (60 images per second) and the temperature resolution of up to 1 K. The accuracy is 0.5% of a measuring value.
For the investigation of the influence of PC on BF streaming conditions (Section 4), the pressure drop was measured in different vertical and horizontal zones using a 2D cold model, Figure 4. Its dimensions are as follows: height─900 mm, width─770 mm, and thickness─210 mm. Compressed air and PC are blown via a tuyere and stream through the packed bed of polyethylene balls with a diameter of 10 mm. The used coal was PC 4 (see Table 2) with grain size of 90-125 μm.
Hot-metal carburization (Section 5.1) was investigated using a 50 kg induction furnace for the melting and casting moulds.
Dripping behavior (Section 5.2) was studied by means of the setup presented in Figure 5. It consists basically of a steel tube where the ignited coke is placed, an injection device for injecting various solids, and a water basin to collect the iron droplets. The setup possesses three tuyeres for the injection of ARA and equipped with a camera and a scale. Thus, it was possible to investigate the effect of injected PC on static and dynamic holdups.

. Coal Conversion and Char Formation
The effect of various factors on the PC conversion in the raceway is summarized in Refs. [7,8]. Trials performed in this study using the injection plant MIRI ( Figure 1) have shown that the reaction temperature and PC feeding rate affect the coal-conversion behavior stronger than other factors. Enriching blast with oxygen increases the conversion degree only slightly, for example, from 50.8% to 52.1% by adding 4% oxygen. This finding is in line with previously reported results that the effect of the local oxygen concentration in lab and pilot plants, as well as at the experimental BF, affects the PC conversion less than that in industrial BFs. [9,16] The main reason for this lies in tuyere design. The share of crosssectional area in which PC reacts with blast compared to the total tuyere cross section in these facilities is higher than that in industrial BFs.
Furthermore, the effect of grain size and its distribution on the coal conversion has been detected by investigations using both a thermogravimetric (TG)/DTA facility and the MIRI rig. Three PC fractions were examined: original size distribution (OSD) used for injection at an industrial BF (62% < 75 μm, 95% < 160 μm) and two sieved fractions: 40-75 and 90-125 μm. The OSD fraction exhibited the best conversion behavior, because within the particle size distribution, there is a share of very fine particles, which react faster and promote the ignition of bigger particles. The grain fraction 90-125 μm showed the lowest reactivity. Thus, the conversion degree in the MIRI trials was 51%, 48%, and 46% for OSD, 40-75 μm and 90-125 μm, respectively, at constant blast temperature (1150°C) and oxygen concentration (26%).
Char collected from the MIRI tests is shown in Figure 6 along with raw PC particles.
Char properties were studied recently in Refs. [14,17]. An important feature of the formation of char from coal can be derived from Table 2. Two key factors, acting in contrary directions, influence the activation energy and, consequently, the reactivity. Volatile matter (VM) content decreases by devolatilization of PC, and reaction surface increases in the same time. Stronger raise in specific surface area compared to the drop in VM causes higher activation energy of char than that of parent coals. This finding was gained using synthetic char, produced from different coals at temperatures of 1300°C in the Tammann furnace drop tube setup (TF-DT), as shown in Figure 2.

Comparative Reaction Behavior of Carbonaceous Residues
As mentioned in the introduction, further carbonaceous residues may exist in the BF and compete with coal char. Table 3 shows grain size, VM, specific surface (BET), and activation energy (E a ) of synthetic chars, CF, and soot (Char production in the TF-DT experimental set is described earlier. CF and soot were produced in scope of a RFCS project Sparerib by Centre de Recherches Métallurgiques (CRM) in the COKARAC furnace and by thyssenkrupp Steel Europe (tkSE) in the laboratory retort respectively).
A comparison of PC and CF conversion in the raceway was examined using the MIRI plant. Blast temperature was kept  constant in all test scenarios at 1150°C. Table 4 shows that the conversion degree of CF is lower than that for CF─PC mixtures. It can be seen by comparison of scenario 1 with scenarios 4 and 5 at same injection rate (700 g h À1 ) and same oxygen enrichment (26%). Comparison of scenarios 1, 2, and 3 shows the better conversion of CF at its smaller amount and higher oxygen concentration in blast.
A comparable consumption of different fractions of PC, char, CF, and soot was also examined using the TG/DTA analysis. Contrary to the MIRI plant, this equipment doesn't simulate the actual BF conditions, but allows for precise comparison of the reaction behavior of different materials at given conditions. Figure 7 shows the results for PC, CF, and soot. To keep the graph readable, various fractions are shown only for PC. The tests were started in argon atmosphere; after reaching the desired temperature of 1400°C, this gas atmosphere was kept further for 15 min, and then switched to air for 45 min (area between vertical  Grain size of parent PC (see Section 3.1). dotted lines in Figure 7), and then by cooling phase switched back to Ar. The mass loss in Ar is primarily related to VM content in the materials. Comparison of three examined grain size fractions, exemplarily shown in Figure 7 for PC, allows for conclusion that the difference in reactivity is rather small, but the original fraction OSD with broader size distribution exhibits somewhat larger mass loss, followed by the smallest fraction (40-75 μm) and then fraction 90-125 μm. It is in line with MIRI results in Section 3.1 and with previous finding. [18] The grain size effect for CF is similar.
The known catalytic effect of alkali on coke reactivity was identified also for char. Tests using the lab rig and procedure described in Ref. [2] (alkali vapor was generated from potassium carbonate placed into the reaction tube of the Tammann furnace, position 9 in Figure 3) were conducted according to the scenario in Table 5. This scenario simulates solution loss reactions in different vertical BF zones. Results have shown that mass losses with alkali load for both char and coke became larger, Figure 8.

Temperature in the Raceway While Injecting PC
Real-gas temperature in the raceway differs from the RAFT due to heat loss, reoxidation of the elements of hot metal dripped through the raceway, rise in radiation while injecting PC, etc. [8] Furthermore, applied measuring techniques measure typically the solid surface temperature. Nevertheless, the effect of coal injection by trend should be the same, that is, temperature should be lower when PC is injected.
Temperature measurements in the raceway zones of two industrial blast furnaces with coal injection were conducted using a TVC introduced in Section 2. The camera itself and additional devices like objectives and filters as well as all settings, an applied software and statistical methods were the same which were used for all the measurements. Therefore, the measurement instruments and analysis techniques had no influence on the measurement results. It is also true for the BF operation parameters like the hot blast characteristics (temperature, moisture, and oxygen enrichment) and production rate, which were kept comparable within the measurement series. The deviation in blast volume was considered by the analysis. Only the PC rate was varied. Figure 9 shows exemplary three images of the temperature distribution at PC rate of 200 and 100 kg t HM

À1
, and when injection was stopped. Processing the results of several measuring campaigns at one BF conducted at three tuyeres indicated an increase in temperature by 35-80°C for different ROIs (regions of interest [3] ) when reducing pulverized coal injection (PCI) from 200 to 100 kg t HM

À1
, and up to 210°C when stopping the injection of 200 kg t HM À1 PC. [18] At another BF, the measured by the TVC temperature became in some cases by 50-80°C higher when stopping the injection of 200 kg t HM À1 PC. Considering the effect of changed blast volume, rather higher drop of the temperature could be expected. In other cases, the measured temperature became, on the contrary, lower by 80-100°C without coal injection. [19] The observed wide range of the temperature change, in some cases even in opposite direction, might be caused, in general, by following reasons: 1) measuring procedure like usage of a camera objective, filter, camera focus, etc., 2) evaluation procedure and choice of ROI and their number, 3) effect of the BF operation parameters such as blast volume, blast temperature, etc., and 4) effect of unburnt coal.
In our case, the effect of measuring and evaluation procedures, and BF operation parameters, can be excluded due to the measures described earlier. The choice of ROI affects the measurement results, essentially, due to varying intensity of coal stream. The effect of unburnt coal particles or char seems to be a significant reason of temperature deviations while high PCI rates.
Unburnt coal or PC conversion degree affects both the measured temperature and RAFT. The decrease of RAFT while  injecting PC and further ARA is usually given per 100 kg of injected substance per tonne of hot metal (see Table 1). It is assumed that an injected ARA is converted in the raceway completely. As showed in Introduction, it is not the case, particularly at high injection rates. Hirsch indicated that unburnt coal particles are heated up and take some heat from the raceway and therefore cause decrease in RAFT. [20] Wu et al. showed that decomposition heat of PC and reduction rate of silica in ash influence the combustion temperature. [21] From our point of view, the rate of char that leaves the raceway may affect the RAFT and the real temperature by following main phenomena: 1) heating up the coal particles resulting in decrease of temperature; 2) release of VM, resulting in the change of temperature, which depends on amount and composition of VM; 3) more coke is gasified in the raceway because more oxygen is remaining due to lower conversion of coal, resulting in the higher temperature.

Char Effect on Gas Permeability
Streaming behavior of coal particles plays a crucial role in both cases─when char will be consumed and when it will be accumulated. In the first case, it is important to know where reactions of secondary gasification or consumption are preferably expected─in the shaft, in the dripping zone, in the hearth, and consequently, which reaction proceeds. In the second case, it is important to know in which BF regions the accumulation of unburnt residues is mostly expected.
With this background, a study on the effect of injected via tuyeres coal fines on the pressure drop and permeability of different BF zones while changing the injection rate and gas flow and velocity was undertaken using a physical cold BF model, Figure 4.
Results showed that the greater the distance from the injection area (raceway), the less particles are accumulated there. Most of the coal particles accumulate close to this zone and along the area, located at the level of the cohesive and dripping zones on the right side of the model, which is in line with the results of mathematical modeling, Figures 10 and 11. In the horizontal direction, a specific pressure drop is the highest at the right side and decreases to the central zones, and even more to the left side, where no injection occurred. It is true even for upper furnace zone (measurement level 7), as shown in Figure 12. With increasing injection rate and gas velocity, the pressure drop increases. [22] Min.  Adapted under the terms of the EUR 29548 license, [3] Copyright 2019, European Commission.  Table 2), three char fractions produced from this PC at 1300°C, and CF (see Table 3) were charged in moulds (50 g) and liquid hot metal (10 kg) was casted there (Figure 13). Carbon content in hot metal was measured by means of the spark emission spectrometry (F-OES) at 17 positions of each ingot. Results presented in Table 6 allow for conclusion that all the tested materials were able to carburize hot metal from initial 4.3% C to the targeted value of 4.6% C. The variations in carburization behavior were as follows. PC and CF show very similar results (mean value 4,61%, deviation 0,051 and 0,053).
Char fractions show somewhat variational behavior. The higher mean value and deviation for the original fraction are attributed, obviously, to its high ash content. The dissolution behavior of narrow and, particularly, smaller fractions is more uniform. It is also in line with the results of Ohno et al. [23]

Dripping Behavior
The dynamic and static holdups in the dripping zone were investigated in the presence of injected carbonaceous residues. The dynamic holdup means liquid drain volume after liquid supply stop, and static holdup means liquid volume stands still on the packed materials. Fifteen experiments including a reference test without injection were carried out with variable parameters such as the injected ARA, the hot-metal temperature, the coke size, the coke bed height, and the injection rate as displayed in Table 7. Coke temperature was kept constant at 1000°C. Figure 14 demonstrates experiments in the setup presented in Figure 5.
After a trial, the setup was disassembled and the filtered dust, the coke pieces, and the hot-metal droplets were taken from three sectioned areas (1/3, 2/3, or 3/3 of the model height) and analyzed. These samples were undergoing chemical, physical, microstructural, mineralogical, and thermographic examinations.
The results show that the different parameters of the setup have an influence on the dripping behavior as well as on the liquid holdups (static and dynamic) as shown in Figure 15. The influence of the parameters can be evaluated as follows.

Coke Particle Size
With a decreasing coke size, the dynamic liquid holdup decreases, while the static liquid holdup increases. This is due to the decreasing voidage with a decreasing particle size. The voidage is 0.43 for a particle size of 30-40 mm and 0.39 for a particle size between 15 and 25 mm. Therefore, with a smaller voidage, less hot metal can drip through the gaps between the coke particles, making it more difficult for the droplets to move through the coke. This leads to a higher number of droplets of hot metal that remain in the model.

Coke Bed Height
With decreasing coke bed height, the dynamic liquid holdup increases and the static liquid holdup decreases. With a lower coke bed height, the hot metal needs to cover less dripping distance between the gaps of the coke particles. At the same time, the calculated residence time of the hot-metal droplets is shortened (from 0.8 s for 180 mm bed height to 0.4 s for 90 mm), which means that fewer hot-metal droplets can solidify in the gaps or on the coke particles in the model, which leads to an increase in the dynamic liquid holdup.

Hot-Metal Temperature
With decreasing hot-metal temperature, the dynamic liquid holdup decreases and the static liquid holdup increases. This is due to the local cooling conditions of the hot-metal droplets.  The higher hot-metal temperature increases the temperature difference between the liquid phase and the first solidified phase. As a result, the cooling process to solidification is slowed and therefore a higher static liquid holdup (more iron droplets solidify in the model on the coke particles) can be observed at a lower hot-metal temperature.

Feeding Rare
With decreasing feeding rate, the dynamic liquid holdup increases while the static liquid holdup decreases. This is due to the fact that the wetting behavior of the coke particles changes with the injection of the ARA. As the injection rate increases, the contact angle between the coke surface and the liquid decreases, which enhances the wetting behavior. This allows more hot metal to accumulate on the coke surface, which then explains a higher static liquid holdup at higher injection rates.

Injected Material
The influence of the ARA on the holdups depends on the chemical composition, the carbon structure as well as the porosity of the carbonaceous materials. Especially the VM and the carbon, sulfur, nitrogen, and oxygen content of the ARA have a direct influence on the holdups. For materials containing VM, Figure 13. Casting moulds (ingots) with carbonaceous materials (above) and with the casted hot metal (below). Table 6. Hot-metal carbon content (wt%) after the dissolution tests. [15] Carbonaceous  their condensation on the coke surface may occur during pyrolysis. This affects the surface tension of the coke and also increases the wettability of the coke surface. Thus, the volatilization of the VM has a direct influence on the wetting behavior of the coke particles. This leads to an increase in the wetting with an increasing amount of VM, resulting in a higher accumulation of more iron droplets on the coke surface, which explains the higher static liquid holdup. In addition, the coke sulfur, nitrogen, oxygen, and carbon contents lower the contact angle and therefore decrease the wettability [24,25] ─explaining why the CF have the lowest static liquid holdup. It should be considered that the examined parameters do not only have an individual effect, but influence the dripping behavior simultaneously.

Effect on Slag Characteristics
To simulate the influence of carbonaceous residues on the BF slags characteristics such as viscosity, melting temperature, and basicity, the software FactSage was used. In addition, laboratory experiments including measurements of viscosity, liquidus and solidus, and basicity were conducted and compared with the calculations.
In Figure 16, the influence of the char particles on the viscosity of the synthetically produced slag ( Table 8) can be seen. The percentage of char is related to the amount of slag. The simulations as well as the experimental viscosity measurements (performed at the German Aerospace Center (DLR)) have shown that the carbonaceous particles increase the viscosity. With increasing char content in the slag, the viscosity increases. This is due to the fact that the slag network (network former and network modifier) is influenced by the char particles. By introducing more network formers (SiO 2 , Al 2 O 3 , P 2 O 5 ) [26] such as from the ash of the char, more networks are formed between the oxides of the slags, so the viscosity increases. If more network formers are dissolved in the slag, the effect of the oxides is therefore more intense. The deviations between simulations and measurements can be explained by the sample preparation, which caused trapped air bubbles in the samples.
Adding char particles results also in a modification in the basicity and, consequently, affects the slag viscosity. This effect is rather small, but it is different for primary, bosh, tuyere, and final slags (F. Perret, Einfluss von kohlenstoffhaltigen Rückständen auf die Prozesse im Hochofenschacht und Unterofen, Doctoral Thesis, RWTH Aachen, forthcoming).
In addition to the chemistry, the physical effect of solid particles on viscosity still has to be considered. As the number of solid particles in the slag increases, the viscosity also increases as the particles do not dissolve and form an obstacle to the liquid slag. At the same time, char particles also have an influence on basicity, which also has an additional impact on viscosity.
Carbonaceous residues also have an influence on the melting behavior of slags, as shown in Figure 17. The melting range can be affected by the direct change of the chemical composition of the slags, as well as by reactions, which take place by dissolving the char particles in the slag. The carbon from the char particles    oxidizes to CO or CO 2 . This allows the oxides in the slag to react. The oxides are reduced and returned to their original oxide form by the redox reaction. This mechanism changes the local chemical composition of the slag and therefore the melting range. Another effect on the melting temperature is caused by the ash components of the char particles. The subsequent reaction of the particles with the oxides of the slag only leaves ash, which changes the chemical composition of the slag as well. Especially the SiO 2 and Al 2 O 3 components change the melting temperature. [23,27,28] Due to their effect, the melting temperature is lowered. At the same time, the char particles in the slag cause a change in the wetting behavior, which also has also an impact on the melting. According to Ref. [29], an increasing contact angle (measured with 0 wt% ¼ 48°-53°, at 1.5 wt% ¼ 53°-57°, at 2.5 wt% char ¼ 54°-61°) decreases the wetting. Local melting points are formed around the solid particles rather than forming a uniformly distributed melt. Only by increasing the melting temperature, the slag can be melted equally.
Nevertheless, it is important that all the reactions take place at the same time and not individually. Therefore, it is possible to see an increase as well as a decrease in the melting range. After all, the effect depends on the amount of char particles in the slag which is also displayed in the simulations and the experiments.

Char in BF Dust and Control of the PC Injection Efficiency
The previously proved fact that unburnt in the raceway coal residues might be consumed by various means leads to the conclusion that the PC injection efficiency should be determined on the basis of its remaining amount in the BF dust. The problem lies in its identification, which is not possible by chemical analysis due to the presence of other carbon sources in the BF dust, primarily, CF, and sometimes also soot. Proposed methods for carbon type differentiation have been analyzed in Ref. [30].
Newly a combined method for identification and quantification of carbon phases in BF dust, based on microscopic examinations and involving thermogravimetric analysis (TGA) technique, has been developed. [30] It consists of three consequent steps.
In the first step, SEM images of BF dust samples are taken and EDX analysis is performed to identify carbon particles and to separate them from other elements in BF dust like iron, silicon, alumina, or manganese oxides. By testing this method, the software ImageJ/Fiji was applied to process the images and to calculate the area distribution of marked particles (Figure 18).
In the second step, a differentiation of char and coke particles is performed by optical examination of particle properties such as grain size, porosity, shape, and sharpness of the edges. These characteristics of carbonaceous particles are determined using a data base which has to be created for synthetic char and coke particles, for example, using light optical microscopy. In such a Figure 17. Effect of char on the melting area of the slag presented in Table 7 (left: FactSage simulation, right: experimental data). Figure 18. SEM image processed by ImageJ. red─C phases, white─iron, grey-Si, Al, Mg. Reproduced with permission [15] Copyright 2021, Verlagshaus Mainz GmbH. way, it is possible to determine the mass ratio of char and coke in the selected area of a polished sample. By testing this method, the assumption, based on preliminary examinations, was made that particles have an average density of 0.9 g cm À3 for char and 1.7 g cm À3 for coke. In an example, presented at Figure 19, the relative shares of char and coke make up 9.6% and 90.4%, respectively, from the total rate of carbonaceous phases, 29.61%. [15] In the third step, BF dust samples are undergoing the TGA analysis. It is based on a method developed by Ng et al., which determined two main reacting areas where reaction of char and coke is expected. [31] This method was modified in terms of temperature scenarios and is supported by a data base of synthetic materials. [29] For creating this data base, different combinations of coal and coke originated synthetic materials undergoing the TGA analysis to receive a calibration. Using calibration results, the ratio of char and coke in any mixture can be determined. Figure 20 shows the mass change rates for two BF dust samples.
The sample BF dust 1 corresponds to a BF operation at PCI rate of 220 kg t HM À1 and coke rate of 310 kg t HM

À1
. The sample BF dust 2 corresponds to a BF operation at PCI rate of 150 kg t HM À1 and coke rate of 350 kg t HM À1 . The mass ratio is calculated as char combustion area divided by the coke combustion area. For BF dust 1, the char/coke mass ratio makes up 1.93, for BF dust 2-0.33.
The elaborated method may contribute to better control of the PC injection efficiency by means of establishing a relationship between the char content in the dust and the PCI and coke rates. As showed earlier, the higher PCI rate is accompanied by the higher amount of char in the BF dust. To achieve reliable quantitative results, an appropriate database is required.

Conclusions
1) The effect of grain size and its distribution on the coal conversion was detected by investigations using the injection rig and TG/DTA facility. The original PC fraction used at an industrial BF showed higher conversion degree than two extracted narrow fractions (40-75 μm and 90-125 μm) because a small share of very fine particles reacts faster and promote the ignition of bigger particles. It is in line with previous finding. [18] The grain size effect for CF is similar. 2) The rate of char that leaves the raceway may affect the RAFT and the real temperature measured by a TVC due to the changes in heat generated/absorbed by heating up the coal particles, release of VM and amount of gasified coke.
3) The known catalytic effect of alkali on coke reactivity was identified also for char by means of the Tammann furnace experimental set. 4) A comparison of PC and CF conversion in the raceway was examined using the MIRI plant. Results show that the conversion degree of CF is lower than that for CF─PC mixtures. 5) Measurements of the pressure drop using a BF cold model showed that most of the injected coal particles accumulate close to the simulated raceway zone and along the area, located at the level of the cohesive and dripping zones. With increasing Figure 19. Characteristic char particles (red marked) and coke particles (black marked), SEM image. Reproduced with permission [15] Copyright 2021, Verlagshaus Mainz GmbH. Figure 20. TGA curves along with temperature profile for blast furnace (BF) dust samples (exemplary for one test series). Reproduced with permission [30] Copyright 2020, DVS Media GmbH. injection rate and gas velocity, the pressure drop increases. 6) Carburization tests showed that all the tested carbonaceous materials were able to carburize hot metal from initial 4.3% C to the targeted value of 4.6% C. 7) In the experimentally simulated dripping zone, the dynamic liquid holdup decreases and the static liquid holdup increases when i) coke size and hot-metal temperature decrease, ii) coke bed height increases, iii) feeding rate of injected carbonaceous materials increases.
The effect of the injected materials type on the holdups depends on their chemical composition, the carbon structure, and the porosity. For example, the volatilization of the VM affects the wetting behavior of the coke particles, resulting in a higher accumulation of more iron droplets on the coke surface, which leads to the higher static liquid holdup. 1) The influence of carbonaceous residues on the BF slags viscosity, melting temperature, and basicity was studied using the software FactSage and experimental measurements. The results testified that with increasing char content in the slag, the viscosity increases. The slag melting behavior can also be affected due to the change of chemistry and wettability. 2) For char identification and quantification in BF dust, a method based on microscopic examinations and involving TGA technique was further developed and tested. It should serve for better control of the PC injection efficiency.