2018 Volume 58 Issue 5 Pages 808-814
In the sintering process during iron making, the sintering reaction proceeds in a packed bed along with the combustion of coke particles. Although detailed temperature information is necessary to improve the process, it is difficult to measure the temperature distribution inside the packed bed with high spatial and time resolution. We performed in situ temperature measurement inside the sinter bed at high spatial and time resolution, i.e., 2 mm and 10 s, respectively, during sintering. A sheathed thermocouple was scanned at optimized scan speed along the inside of the thin-wall alumina tube which was held perpendicular inside the sinter bed. The information on the temperature variation during sintering showed a clear correlation between the quality of the sinter and the sinter heat pattern for each layer. Further analysis also showed that the flame front speed is proportional to the O2 consumption in the sinter bed. The temperature measurement technique enabled an unprecedented detailed discussion with the temperature distribution inside the sinter bed during sintering. This technique will not only help to improve the sintering process but also provide beneficial information on the chemical reactions occurring inside packed beds.
The sintering process for iron making has been used to produce sinter from granular iron ore and other raw materials. As the price of iron ore rises over the long term, the steel industry is increasing the use of inexpensive finer iron ore in the iron ore sintering process. Such market trends enhance the importance of the sintering process.
It is known that not only the size of the sinter but also its reducibility is altered through the sintering process. Considerable research has been devoted to clarify the relation between the sintering process conditions and reducibility of the achieved pellets from the process.1,2,3,4,5,6,7) These studies showed that sinter yield and quality are influenced by many parameters of the process such as the chemical composition of raw materials,1,2,3,4,5) composition of flowing gas,3,5,6,8,9) and the granulation method.10,11) Among all parameters, the sintering heat pattern in the sinter beds significantly affects the sinter from the viewpoint of reducibility as well as yields.12,13) In situ X-ray diffraction (XRD) results showed that compound oxides drastically transform as the temperature increases from room temperature.4,5,6,14) In situ XRD studies showed that CaO and Fe2O3 start reacting to form mono-calcium ferrite (CF) at around 1000°C. The amount of CF gradually increased as the temperature increased. At around 1100°C, acicular silico-ferrite of calcium and aluminum I type (SFCA-I) starts growing. SFCA-I is famous for its high reducibility and mechanical stability during the reduction process. However, SFCA-I is also known to decompose into a liquid phase and iron oxides at around 1200°C. Over 1300°C, SiO2 content in liquid phase increases. Then, columnar silico-ferrite of calcium and aluminum (SFCA) forms with the decreasing temperature of the sinter bed. SFCA-I remains in the sinter in the case of a lower peak temperature.15) SFCA is inferior to SFCA-I in terms of the reducibility and reduction degradation index. Therefore, it is important to control the peak temperature in the sinter bed in order to have more SFCA-I content. Next, we consider the heating rate of the sinter mixture in the sinter bed. The temperature inside the real sinter bed soars so fast on the flame front. The heating rate at the flame front reaches more than 1000°C/min. Chemical reaction at such a high heating rate is different from that of semi-static reaction that is usually used for in situ XRD studies. Due to the difficulty in in situ observations at the high heating rate, there are still many debates on the formation mechanism of SFCA-I and other forms of calcium ferrites.4,5,6,7,14,15,16) As shown above, the temperature in the sinter bed has a strong impact on the quality and yields of the sinter.
Many engineers and researchers have been working on how to control the heat patterns in the sinter bed in order to increase the yields as well as quality of the sinter. Several technologies are widely applied to the sintering process. Segregation control is one such technology.11,17,18) The high-temperature hold time tends to be short in the upper layer of the sinter bed, while the peak temperature of the lower layer is prone to be too high. Segregating coke particles into the top surface moderates the temperature at all layers and improves the quality of the sinter.
As discussed above, it is crucial to tune the heat pattern inside the sinter bed in order to improve the quality of the sintered pellets and yields. Detailed information about the temperature in the sinter bed is necessary for further improvement of the sintering process. However, there are several difficulties in measuring the temperature in the sinter bed. The first is how to insert the sensors. Iron ore strongly adsorbs a wide range of electromagnetic waves. In addition, the temperature rises to as high as 1300°C. These experimental difficulties limit the measuring methods to thermocouples. However, inserting thermocouples crashes the packed bed and perturbs the gas flow in the sinter bed. Therefore, these accompanying issues limit the number of thermocouples inserted into the sinter bed. In many cases, only two or three thermocouples are inserted into the sinter beds even in the laboratory tests.
The second difficulty is the spatial variation in the temperature. The sinter bed is a mixture of raw materials: iron ore, limestone, olivine, coal, quicklime, hydrated lime, serpentine, and others.19) Among all raw materials, only coal is combustible. Only 3 to 5 wt% of coal particles are sparsely dispersed inside the sinter bed. Combustion heat of each coal particle heats the sinter mixture existing around it. Therefore, temperature varies depending on the distance from the coal particles. This results in the inhomogeneity of the temperature at a millimeter scale. In addition, it is known that the temperature of the sinter bed decreases around a wall of the pallets. This causes temperature variation at a centimeter scale. All temperature data are under the influence of the temperature inhomogeneity at a millimeter and a centimeter scale. Inhomogeneity of temperature inside the sinter bed complicates the interpretation of the data.
The third difficulty is the moving flame front. The flame front of the sinter beds is propelled downward following the downward air flow. Moreover, it is well known that the temperature of the lower part of the sinter bed is higher than that of the upper part. This temperature variation causes a variation in the quality and yields of the sintered pellets along the depth of the sinter beds. Therefore, it is desirable to measure the temperature during combustion at all depths.
As discussed above, measuring the temperature inside packed beds is a complicated but important issue. It is desirable to measure the temperature at many points without perturbing the gas flow while the chemical reactions are taking place. This is also true even in the field of catalytic reaction. Jamal Touitou et al. reported several techniques of in situ monitoring of catalyst beds.20,21) They used a glass tube equipped with a fine thermocouple inside of it. The glass tube was inserted through the packed bed parallel to the reaction gas flow. The temperature inside the packed bed was determined during the reaction by sliding the packed bed itself. This technique unveiled the temperature variation inside the catalyst bed at a spatial resolution of 0.5 mm. Although this technique is apparently feasible even in the sinter bed, there are important differences between the catalyst beds and the sinter bed. The first difference is the dynamics of the reaction. The catalytic reactions basically proceed as steady state reactions in catalyst beds. The reaction rate and temperature distribution in the catalyst bed varies slowly due to the long term deactivation of the catalyst. Therefore, it is not a problem even if it takes so long time to measure the temperature of each place inside the catalyst bed in order to determine the temperature distribution. In contrast, the sintering reaction is dynamic; the temperature varies with time. It is necessary to measure the temperature in a short period of time. The second difference is the temperature. The sinter bed can reach a temperature of more than 1300°C. The glass tubes melt at such high temperatures. The third difference is the formation of the melt. In the sinter bed, melted CF is formed. This melt sticks to the tubes and makes it difficult to move the tubes. In this study, we overcame the difficulties caused by the differences and performed in situ temperature measurement in the sinter beds at high spatial and time resolutions.
We used a sinter pot with an inner diameter of 100 mm and depth of 440 mm. Glass-wool insulation sheets were placed on the inner wall of the sinter pot. The glass-wool insulation was 10-mm thick, and the empty space for filling sinter mixture remained with 80 mm in a diameter. An alumina tube (TRIO ceramics, PTO series), which has an inner diameter of 1.2 mm and outer diameter of 1.8 mm and length of 460 mm, was inserted along the central axis of the sinter pot. The thickness of the tube’s wall is 0.3 mm. The diameter of the tube was smaller than the average diameter of the raw materials. Therefore, we assume that the effect of the wall of the alumina tube during the sintering test was small. The alumina tube was held perpendicular to the top surface of the sinter pot with a piece of string. The upper-side hole of the alumina tube was covered by a piece of scotch tape in order to prevent clogging. Raw materials were thrown into the sinter pot to fully fill the vacant space. The composition of the sinter mixture is listed in Table 1. The raw materials were mixed using a granulating machine before they were thrown into the sinter pot. The scotch tape attached on the hole was removed. To verify the reliability of data obtained from the thermocouple in the alumina tube, type R thermocouples were inserted into the sinter bed through the holes opened on the sidewall of the sinter pot. The holes are 4 mm in diameter and located at 240 and 360 mm from the top surface. After initiating an air flow with a negative pressure of 80 mmAq, the top surface of the raw materials was ignited for a minute with the burner fueled by natural gas. The gas composition of the exhaust gas was maintained and recorded with a gas analyzer (HORIBA, PG-250 portable gas analyzer) until the end of the sintering process.
| Raw material | Ratio (wt%) |
|---|---|
| Iron ore | 69.3 |
| Lime stone | 11 |
| Olivine | 2.55 |
| Quicklime | 0.84 |
| Return Ore | 12.6 |
| Coke | 3.77 |
| Total | 100 |
On completion of the ignition, an electric linear slider (Oriental motor, EAS series) equipped with a step motor was placed on the sinter bed. A high temperature tolerant type-K sheathed thermocouple (Okazaki manufacturing, HOSKINS2300 series, outer diameter: 1.0 mm, Sheath length: 500 mm) was inserted into the alumina tube. A thermocouple of 1.0 mm in its diameter was used to reduce the heat capacity of the thermocouple and to increase the temperature-following speed. A thermocouple of 2.0 mm in its diameter was also subjected to the same experiments shown below, but the thick thermocouple showed lower-temperature following speed (data not shown). By contrast, too thin thermocouple leads the mechanical weakness and the thermal shunting error during the measurement. Considering the all features shown above, we concluded that the diameter of 1.0 mm is the best option for the experiments of our study. Pt-Rh sheathed R-type thermocouples would be better choice if they are available, but we used K-type thermocouple due to the budget reason. A sleeve of the thermocouple was fixed on a manipulator of the slider in order to control the thermocouple. Monitored temperature values were recorded by a data logger (T&D, MCR-4TC) every tenth of a second. We note the importance of the gap between the inner wall of the alumina tube and the thermocouple. It is preferable to narrow the gaps from the viewpoint of heat transfer between the sinter bed and the thermocouple. A narrower gap ensures higher heat transfer and small differences between the monitored temperature and the real temperature of the sinter bed. In contrast, if the gaps are too narrow, the thermocouples cannot move inside the alumina tube due to frictional resistance. It is necessary to ensure that the gap between the thermocouple and the inner wall of the alumina tube is ideal before starting the experiment. The thermocouple was being scanned by the electric slider during the sintering test. The scan rate was 3.0 mm/s for the upward direction and 5.0 mm/s for the downward direction. The air flow rate during sintering was controlled by the negative pressure below the sinter bed. The negative pressure at the bottom of the sinter bed was kept at 350 mmAq throughout the sintering. Schematic of the pot test equipment during sintering is shown in Fig. 1. The thermocouple was scanned until the completion of the sinter test.
Schematic of the pot test equipment for in situ temperature measurement. (Online version in color.)
After having completed the sintering process, the obtained sinter was divided into seven equal parts from the top to the bottom. Each portion was subjected to the shutter test and sieve classification.
The raw materials in the sinter bed undergo chemical reactions to form sinter during the sintering process. CF appear at temperatures higher than 1100°C.4,5,6,14) It binds the raw materials in the sinter bed to form sinter. Important chemical reactions proceed as a result of the heat produced by the combustion of coke particles. Therefore, high-temperature holding time has been an important index to assess the quality of the sinter. In contrast, the temperature before the ignition of the coke particles is not important for the discussion. Therefore, we scanned the thermocouple upward and downward in the alumina tube only in the vicinity of the combusting layer. Figure 2 shows the location of the thermocouple tip at each sintering time. There was no data in the first 100 s because placing the electric slider and the thermocouple required approximately 1 min. The time variation of the monitored temperature is shown in Fig. 3. Many sharp peaks were observed in Fig. 3. Every upward or downward scan in Fig. 2 went through the combusting layer, which appeared as a peak in Fig. 3. The scan speeds, 5.0 mm/s for the downward scan and 3.0 mm/s for the upward scan, were confirmed to exhibit small differences between the temperature values measured by the scanned thermocouple and the two static thermocouples (Fig. 4). Small difference between the values was ascribed to the distance between the static thermocouple and the scanned thermocouple; the static thermocouples were located more than 4-mm away from the thin alumina tube in order to avoid the collision and breaking of the thin alumina tube for the scanned thermocouple. The similarities of the peak temperature value indicate that scanned K-type thermocouple and static R-type thermocouple showed similar value even at higher temperature than 1250°C, which is normally upper-bound suitable temperature for K-type thermocouple. We assumed that short time exposure to the higher temperature than 1250°C didn’t give severe damage on the K-type thermocouple because the melting point of K-type thermocouple is around 1400°C. We determined the scanning speed considering three points. The first point is the time required for the heat transfer. It is necessary to scan slowly enough to transfer the heat of the sinter bed to the thermocouple inside the alumina tube. The second point is the time required for each scan. The faster is the scan speed, the shorter is the time required for each scan, which realizes the higher time resolution. The third point is a traveling direction of the flame front. The flame front moves downward during sintering. Generally, the flame front speed (FFS) is in the range from 15 to 30 mm/min, which is identical that in the range from 0.25 to 0.5 mm/s. This means that the downward scan takes a longer time to catch up with the flame front than the upward scan if the scanning speed is constant. Therefore, it is required to move the thermocouple faster in the downward direction than in the upward direction. The data acquired at the other scan speed than that for Fig. 5 is shown in the Appendix A. A scan speed of 6.0 mm/s gave higher temperature values for the downward scan than those for the upward scan (Table A.1, Fig. A.1), which indicates that scan speed of 6.0 mm/s is too high for the upward scan. On the other hand, the scan speed of 4.0 mm/s for both the directions lowers the time resolution (Fig. A.2).
Time variation of the thermocouple-tip position during sintering.
Time variation of the monitored temperature during sintering.
Temperature values measured by scanned thermocouple (TC) and static TC at the same position.
Time variation of the temperature distribution (a) and the contour (b).
The relation between the sintering time and the monitored place is illustrated in Fig. 2, while the time variation of the monitored temperature is shown in Fig. 3. Therefore, combination of data held in Figs. 2 and 3 makes a three-dimensional graph that shows the temperature at each place and time. Figures 5(a) and 5(b) show the temperature contour calculated from the data of Figs. 2 and 3. The interval between data points were 10 s along the time axis and 2 mm along the place axis. Each data point was obtained by interpolating the obtained data in Figs. 2 and 3. All the calculations were performed with MATLAB. Figure 5(a) shows that the lower layers have higher peak temperatures. In addition, we can understand from Fig. 5(b) that the lower layers have a longer high-temperature hold time. These results are consistent with the widely accepted tendency in the sinter bed.11,12,13) Some computer simulation results also showed counterparts of Fig. 5(a).22,23,24) However, to date, no reports have experimentally determined the time variation of the temperature at such a high time and spatial resolution, i.e., 10 s and 2 mm, respectively. It is worth noting that each sinter pot test gives distinctly different counterparts of Fig. 5(a) even if the experiment is performed under the same experimental condition. A couple of examples are shown in the Appendix A (Figs. A.1 and A.2). These differences were speculated to be caused by the small differences in various parameters including the homogeneity of the raw materials, ambient temperature, relative humidity, and filling rate. The unprecedented data shown in Figs. 5(a) and 5(b) enabled to clarify the effect of many parameters on the heat patterns inside the sinter bed.
3.3. High-temperature Holding Time and Quality of the SinterThe scanning temperature measurement unveiled the temperature variation in the sinter bed at high resolution of time and space. We performed further analysis using the data shown in Figs. 5(a) and 5(b). Figure 5(a) illustrates the temperature history of each place at the spatial resolution of 2 mm. We compared the temperature history of each layer and the quality of the sinter at the same position. We equally divided all the obtained sinter into seven parts from the top to the bottom and calculated the average high-temperature hold time for the corresponding layers. As for the top portion, we did not consider the temperature data of the top 20 mm when we calculated the high-temperature hold time because no temperature data was available before the temperature of that region reached its peak. Each portion was separately subjected to the shutter test and sieve classification. It is worth noting that the alumina tube was partly covered by the solidified melt, which indicates that wall effect was small.
Figure 6 shows the average holding time of temperature above 1100°C, which is around the starting point of the formation of SFCA,5,6) and the grain size distribution after the shutter test for each portion. As for the top three portions in Fig. 6, both the high-temperature hold time and the ratio of the larger pellets than 5 mm increase as the layer goes down. This indicates that the amount of the melt to form sinter increased as the high-temperature hold time increased. In contrast, once the high-temperature hold time reached about 40 s, the ratio of the pellets larger than 5 mm got saturated. This result indicates that there is a threshold of the high-temperature hold time above which it has less effect on the yields. Next we note that the bottom portion has the large amount of coarse particles. The bottom portion has a high-temperature hold time that is twice that of the middle portions. These results indicate that the longer high-temperature hold time caused more melt and let the sinter grow larger. The clear relation between the temperature history and quality of the sinter was confirmed by the analysis of the data shown in Figs. 5(a) and 5(b).
Average holding time of temperature above 1100°C, and the grain size distribution after the shutter test for each portion of the sinter.
We further analyzed the data shown in Fig. 5(b) in order to discuss the propagation of the combustion. All the pot tests used cokes, which start combustion at around 600°C (Fig. A.3). Therefore, the lower line in Fig. 5(b) corresponding to 600°C can be assumed as the flame front line of the sinter bed. We calculated the FFS at each position by numerically differentiating the flame front line with respect to the sintering time. Numerically differentiated values were calculated as the slope of the approximate straight line for 60 s before and after each time. The calculated FFS values are illustrated in Fig. 7. FFS exhibited a tendency to increase as the flame front went down. This tendency is well known and reported in many other reports.12,13,22,23,24,25) Some computer simulation results are in good agreement with the data of Fig. 7.22,23,24) However, there are no reports that experimentally determine FFS values at each depth of the sinter bed at such a high spatial resolution.
Time variation of the flame front speed (FFS) and that of the O2 consumption in the sinter bed.
To verify the validity of the FFS values in Fig. 7, we calculated the amount of O2 consumption at each time using the gas composition and flow rate of the exhaust gas. As the flame front proceeds, particles of the cokes located at the flame front get combusted. Therefore, the amount of O2 consumption correlates with FFS, as shown in the next equation, where CA is the amount of the carbon in unit volume of the raw material, FFS is the flame front speed, RA is the rate of combustion, and NO2 is the amount of O2 consumption due to the combustion.
| (1) |
Assuming the homogeneity of the sinter bed subjected to the pot test, CA and RA in Eq. (1) are constant values. Therefore, NO2 would linearly correlate with FFS.
The time variation of NO2 is plotted in Fig. 7 along with FFS. Aside from the deviation from each other in the first 300 s, there were only small differences between the time trends of FFS and NO2. This result strongly indicates the validity of FFS values and the method to determine the FFS values. Next we discuss the origin of the difference in the time trend between FFS and NO2, especially the first 300 s. The sintering time of the first 300 s corresponds to the top 100 mm. It is widely known that the top layers of the sinter beds are prone to have low yields. One cause of this phenomenon is the imperfect combustion of cokes in the top layer. This trend leads lower RA values than in the other layers. In order to balance the both side of Eq. (1), it is necessary to increase the ratio of FFS to NO2 when RA is small. This results in higher FFS in the top layer. Another possible explanation for the deviation is the packing density. The top layer tends to have lower packing density due to the lack of static pressure from the layer above it. The lower packing density causes a lower CA value and higher ratio of FFS to NO2. Similar studies have been conducted using average FFS during sintering.25,26) However, as shown above, the time trends of FFS and NO2 clearly enable the detailed discussion about the relation between coke combustion and bed condition at each layer. This information will make it possible to discuss the effect of segregation on the combustion and temperature in the sinter bed.
The validity of the data shown in Figs. 5(a) and 5(b) is evident considering the clear relation between FFS and NO2 in Fig. 7. Here we consider the origin of time variation in FFS. It is known that the ignition is led by the heat flow from the layer above the igniting layer. We calculated several values related to the heat flow and found that FFS varies proportionately to the thickness of the layer at a higher temperature than the ignition temperature of coke, which is 600°C. Figure 8 illustrate the results. This relation is rationalized by the efficiency of the heat flow from upstream to downstream. FFS would increase if the igniting layer is efficiently heated by the heat flow from the upstream. The heat of the sinter bed is transferred from the solid phase to the gaseous phase in the sinter bed. Then, the gaseous phase flows downward to give heat to the downstream layer that is igniting. This implies that the amount of heat transfer from the solid phase to the gaseous phase is crucial to determine the FFS. As the thickness of the layer at a temperature higher than 600°C increased, the heat transfer from the solid to the gaseous phase would be more efficient. As a result, the downstream layer would get heated in a shorter period of time, which results in the higher FFS.
Time variation of the flame front speed (FFS) and that of the thickness of the layer above 600°C.
We analyzed the temperature inside the sinter bed at high spatial and time resolution, i.e., 2 mm and 10 s, respectively, during sintering. A sheathed thermocouple was scanned along the inside of the alumina tube, which was held perpendicular inside the sinter bed. Alumina tubes having a thin wall and optimized scan speeds realized the in situ measurement of temperature. The information on the temperature variation during sintering enabled detailed discussions about the relation between the quality of the sinter and temperature history at each layer. Analyzing the sinter equally divided in the layer thickness direction, grain size distribution of the sinter after the shutter test was confirmed to relate to the high-temperature holding time of each layer. Further analysis showed that FFS is proportional to the O2 consumption in the sinter bed as well as the thickness of the layer at temperatures above 600°C. The temperature measurement technique enabled an unprecedented detailed discussion about the sintering process. This technique will help to improve the sintering process of the steel industry.
Time variation of the temperature distribution measured by the scan-rate of 4 mm/s. Large waviness was observed on the flame front at sintering time of 1200 to 1400 sec. Due to the slow scan speed, time span required for each scan increased. Large time span between the data points caused the waviness, because the interpolation between the data point couldn’t be performed correctly.
Coke combustion measured by Thermo gravimetric analysis equipped with mass spectroscopy (TGA-MS). CO2 (FW = 44) was used for monitoring. The amount of coke was 10 mg. The sample was combusted under the air flow of 50 ml/min. The heating rate was 50°C.
| Scan number | Up (°C) | Down (°C) |
|---|---|---|
| 1 | 945.2 | 1077.4 |
| 2 | 989.4 | 1148.7 |
| 3 | 1106.2 | 1224.1 |
| 4 | 1061.5 | 1157.8 |
| 5 | 1034.7 | 1201.3 |
| 6 | 1116.7 | 1206.2 |
| 7 | 1135.5 | 1212.1 |
| 8 | 1120.5 | 1211.8 |
| 9 | 1101.8 | 1241.8 |
| 10 | 1108.6 | 1214.9 |
| 11 | 1234.1 | 1270.4 |
| 12 | 1266.2 | 1309.7 |
| 13 | 1249.7 | 1355.9 |
| 14 | 1242.9 | 1316.7 |
| 15 | 1242.9 | 1332.3 |
| 16 | 1246.5 | 1253.4 |
| 17 | 1177.3 | 1260.7 |
| 18 | 1217.3 | 1263.8 |
| 19 | 1187.0 | 1226.3 |
| 20 | 1189.1 | 1230.9 |
| 21 | 1195.0 | 1237 |
The high scan speed caused the large difference in detected temperatures between the scans in upward and downward direction.
