2016 年 56 巻 11 号 p. 1948-1955
Coke strength is an important property and generally evaluated with Drum Index (DI). It is considered that coke breakage occurs by two types of breakages during DI measurement, surface breakage and volume breakage. In this study, the cause of volume breakage was investigated by image analysis of cross-sectional image of coke before and after the DI measurement. As a result, a new image analysis method was developed. In this image analysis method, pore shape and coke-matrix connectivity were evaluated and new parameters “degree of Irregularity (I)” and “Connectivity Index (CI)” were proposed. I and CI represent pore shape and coke-matrix connectivity, respectively. By using this image analysis method, it was revealed that pore shape and coke-matrix connectivity were improved on adding HPC (High Performance Caking additive). By comparing the cross-sections before and after the DI measurement, it was found that the volume breakage occurred at the portions which have higher porosity, coarse pores, cracks and thin pore wall. Furthermore, it was clarified that the thickness of the pore wall broken by the DI measurement was < 7.6 µm thickness.
Role of coke is not only to act as a heat source in the blast furnace, reducing agent for iron ore, as a spacer for liquid permeability in the blast furnace, but also to provide the mechanical strength to hold the reactor bed. Coke strength, therefore, is a very important property. Industrially, coke strength is evaluated with DI (Drum Index: Weight percentage of coke remaining on the sieve of certain mesh after rotating the coke drum for given number of revolutions). It is considered that coke breakage occurs by two types of breakages during DI measurement, surface breakage and volume breakage.1,2,3,4)
Several literatures on the surface breakage have been reported. Uebo et al.5) measured the porosity, the Vickers hardness and the Brinell hardness of cokes and compared these properties. The Vickers hardness and the Brinell hardness stand for the degree of hardness of coke-matrix and of coke surface hardness including the effect of pore structure, respectively. Thus, the Brinell hardness is influenced by the porosity rather than coke-matrix hardness. Arima6) compared the porosity and DI1506 (weight percentage of coke remaining on the sieve of 6 mm after rotating the coke drum for 150 revolutions) and found that DI1506 increased with decreasing porosity. Furthermore, Arima6) took images of coke particles which were generated from micro-strength index (MSI: coke surface hardness) measurement and investigated the relationship between the porosity of coke particles and MSI. Increase in the porosity of coke particles decreased the MSI. Furthermore, it was reported that the defects in non-adhesion grain boundaries and large pores in the coke particles made a significant influence on the MSI. From the above reports, it is considered that the main factor of the surface breakage can be the defects which are high porosity, non-adhesion grain boundaries, large pores and connected pores, rather than the coke-matrix hardness.
On the other hand, there are a few studies touching on the volume breakage.2,3,4,5) While other causes are also not fully understood, it is considered that one of the main reasons for volume breakage is the presence of cracks in the cokes.2,3,4,5) It was reported that 6–30% of coke particles generated by DI15015 measurement, was due to volume breakage.1,7) Thus, it is important to clarify the cause of the volume breakage.2)
In the present work, influence of DI15015 measurement on cokes and the mechanism of the volume breakage were investigated by comparing the pore and the coke-matrix structures in the cokes before and after DI measurements.
Three coals (Coal A, Coal B and Coal C) and one caking additive (HPC, High Performance Caking additive) were used. Coal A and B were caking coals, and Coal C was a slightly caking coal. HPC was produced from a steam coal by solvent extraction at 653 K.8) The results of the ultimate analysis and the proximate analysis and logarithm of maximum fluidity of coals and HPC are shown in Table 1. Two different coke samples (Coke A and Coke B) were prepared by blending these coals and additive. Each blending ratio and the degree of DI15015 are shown in Table 2.
| Sample | Proximate analysis | Ultimate analysis | logMF [–] | |||||
|---|---|---|---|---|---|---|---|---|
| ash | VM | C | H | N | S | Odiff. | ||
| [wt%] (dry basis) | [wt%] (daf basis) | |||||||
| Coal A | 14.1 | 18.7 | 90.6 | 4.8 | 2.4 | 0.7 | 1.4 | 1.61 |
| Coal B | 9.9 | 21.6 | 89.7 | 4.9 | 1.7 | 0.4 | 3.4 | 2.40 |
| Coal C | 9.6 | 25.7 | 86.6 | 5.0 | 2.0 | 0.6 | 5.8 | 0.95 |
| HPC | 0.8 | 44.2 | 87.0 | 5.6 | 1.9 | 0.6 | 4.2 | 4.78 |
| Bulk density [kg/m3] | Particle size | Coal blend [%] | DI15015 | ||||
|---|---|---|---|---|---|---|---|
| Caking coal Coal A | Caking coal Coal B | Slightly caking coal Coal C | Caking additive HPC | ||||
| Coke A | 800 | <3 mm (80%) | 20 | 30 | 50 | 0 | 77.8 |
| Coke B | 20 | 25 | 50 | 5 | 82.7 | ||
Carbonization method to produce a coke sample is as follows. Coals blended at the ratio as indicated in Table 2 were carbonized in a steel box (L: 233 mm × W: 233 mm × H: 245 mm) placed in an electric furnace at a heating rate of 3 K/min (achieved temperature was 1323 K) while flowing nitrogen gas. During heating, the steel box was surrounded with the coke breeze. When the temperature at the center of the coke lump reached 1223 K, it was held for 2 hours at this temperature. After carbonization was completed, the steel box and coke breeze were taken out from the electric furnace and put in a sealed container and cooled down to the room temperature by keeping overnight.
2.2. Resin EmbeddingSchematic of resin embedding procedure is shown in Fig. 1. The epoxy resin used for resin embedding is Herzog epo (low viscosity type, HERZOG Co., Ltd.). Coke, casted under 25.4 mm, was placed into a cup of 25.4 mm diameter and filled with the resin under the reduced pressure (100 hPa) and were left for 2 days at room temperature for hardening of resin. Samples were then dried and cut at the center and divided into two pieces. The surface was polished with 165 μm abrasive paper, resin was poured on the polished surface under 100 hPa to impregnate it perfectly and leave it for 5 days at room temperature for hardening of the resin. After drying, the surface was polished again and then photographed. This procedure was done for three coke pieces of each coal blend. As a result, six surfaces for each sample were photographed.
Schematic of resin embedding procedure.
Images were taken with a digital microscope VHX-600 (KEYENCE corporation, object lens: VH-Z20), image field is 6.1×4.6 mm and 1600×1200 pixels size (magnification: 50 times, resolution: 3.8 μm/pixel). For each sample, 30 images were taken and these images were composed using image processing software ICE (Image Composite Editor, Microsoft corporation), to get a panoramic view of 5000×5000 pixels resolution. The procedure is shown in Fig. 2.
Schematic of taking cross-sectional panorama image of coke piece.
All the image processing was performed using image processing software ImageJ (Wayne Rasband).
2.4.1. Image BinarizationThe panoramic images were first binarized to perform the image analysis. The brightness histogram of the panoramic image is shown in Fig. 3. In the panoramic image, pores, resin, coke-matrix and clear cup (for resin embedding) can be seen. However, the brightness histograms of these parts were partially overlapped. To eliminate the arbitrariness, the bottom of the peaks of resin and coke-matrix was defined as the cut-off threshold (Mode method). If the brightness of pixels in the panoramic image is higher or lower than threshold, the pixels were defined as white or black, respectively. By this processing, almost all of the coke-matrix became white and other parts became black.
Brightness histogram of panoramic image.
For pore structure analysis, sample periphery must be removed and was done by cutting out a square image (2000×2000~3000×3000 pixels) from the binarized panoramic image. An example of square image used for the pore structure analysis is shown in Fig. 4. The perimeter l and the pore area S of all pores in the square image were measured. Within the pore, there was isolated coke-matrix (Non-connected coke-matrix), as shown in Fig. 5(a). This coke-matrix does not influence the feature of the pores. It was assumed that the isolated coke-matrixes were part of the pore as shown in Fig. 5(b). Therefore, S was the pore area that includes the area of isolated coke-matrix within the pore. The circularity (roundness) was calculated from l and S by using following equation.
| (1) |
An example of square image used for pore structure analysis (The coke A before the DI measurement).
Pore structure (a) with and (b) without isolated coke-matrix.
In the present work, we proposed a new parameter for pore structure representation based on the method of Kubota et al.11) Kubota et al. measured the pore structure of 9 single-coal cokes and 4 blend-coal cokes with DI1506 from 83.7–90.0.11) They reported that there was a correlation between DI1506 and the total perimeter of low roundness pores (R < 0.2) in the unit area. It was estimated that the coke with longer perimeter had more originating points that lead to fracture. Also, there was a correlation between the perimeter and the pore area in the coke. The relationship between the perimeter and the pore area in the coke A before the DI measurement is shown in the Fig. 6. The perimeter increased with increasing pore area. The hardness of the materials decreases with increasing pore area because the stress intensity factor increases with increasing pore area. Thus, the stress intensity inside the coke should also be considered by using perimeter. The other feature in the Kubota’s method is “Not considering the high circularity pores”. It was considered that the pore circularity is related to the stress concentration factor. Since the stress concentration factor decreases with increasing circularity of pores, the effect of high circularity pores on the material hardness would be small. Therefore, the high circularity pores were not considered in the Kubota’s method.
The relationship between the perimeter and the pore area in the coke A before the DI measurement.
On the other hand, the stress concentration factor increases with decreasing circularity of pores. In Kubota’s method, the effect of low circularity pores was underestimated. In this work, a new parameter “degree of Irregularity ( I )” represented by following equation was proposed.
| (2) |
Itotal of coke and the total perimeter of low roundness pores of coke before the DI measurement.
In the present work, the pore structure was evaluated by not only the pore shape but also the pore wall thickness. The pore wall thickness was evaluated by the coke-matrix connectivity of the coke cross-sectional image. The image processing procedure to evaluate the coke-matrix connectivity is shown in Fig. 8(a). In the figure, the left image is a schematic of binarized coke cross-sectional image. White color is coke-matrix and black color is pore and background. First, area fraction of all coke-matrix which were isolated from the largest coke-matrix was counted and defined as “Non-connected coke-matrix” (black small dot color). Next, as shown in Fig. 8(a) (1st step), coke-matrix was sharpened 1 pixel (3.8 μm) depth and becomes thinner by total 2 pixels (7.6 μm). Then the coke-matrix was thickened 1 pixel (3.8 μm) and becomes thicker by total 2 pixels (7.6 μm). In this step, coke-matrix thicker than 7.6 μm does not change, but thinner than 7.6 μm disappeared. The disappeared thin coke-matrix (thickness under 7.6 μm) at this step is defined as “2 pixels wall (7.6 μm wall)”. Coke-matrix isolated from the coke-matrix with the largest area fraction by removing 7.6 μm wall is defined as “2 pixel connection (7.6 μm connection)”. In the same way, as shown in Fig. 8(a) (2nd step), coke-matrix was sharpened 2 pixels (7.6 μm) depth and becomes thinner by total 4 pixels (15.2 μm). Then the coke-matrix was thickened 2 pixels (7.6 μm) and becomes thicker by total 4 pixels (15.2 μm). In this step, coke-matrix thicker than 15.2 μm does not change, but thinner than 15.2 μm was disappeared. At this point, disappeared coke-matrix (thickness under 15.2 μm and over 7.6 μm) is defined as “4 pixels wall (15.2 μm wall)”. Coke-matrix which has the largest area fraction by removing 15.2 μm wall is defined as “over 5 pixel connection (over 19 μm connection)”. Then, except 7.6 μm connection, coke-matrix isolated from over 19 μm connection is defined as “4 pixel connection (15.2 μm connection)”. By repeatedly performing the above operations, all coke-matrix was colored accordingly as shown in Fig. 8(b). Area fraction of the colored coke-matrix is shown in Fig. 8(c).
Schematic of image processing; (a) Procedure of coke-matrix connectivity analysis (b) Colored coke-matrix by image processing (c) Area fraction of colored coke-matrix.
The visual appearance of coke B before and after the DI measurements is shown in Fig. 9. The coke before the DI measurement had rough surface and angular shape. While, after the DI measurement, the coke surface and shape became smooth and round. This trend was also observed for coke A. The reason for this change could be the surface breakage during the DI measurement. In addition, it is possible that the volume breakage have also occurred during the DI measurement. However, it is difficult to differentiate and determine where the volume breakage had happened.
The visual appearance of coke B before and after the DI measurement.
The average porosity of cross-section of coke A and B before and after the DI measurement are shown in Fig. 10. The average porosity of both cokes was decreased by the DI measurement. It means that the high porosity portions were removed by the DI measurement. In case of coke A and B, the high porosity potions were distributed in the coke inside. The high porosity portions in the coke A before the DI measurement were shown in Fig. 11. In this figure, the porosity of this cross-section was 53.7% and gray color means that the porosity of 200 × 200 pixels (0.76 × 0.76 mm) block was over 58.7%. If the breakage behavior of the DI measurement was only the surface breakage, the low porosity portions distributed in outside of coke would be removed and the porosity would be higher. Therefore, it was considered that the volume breakage occurs at the high porosity portion, and mainly the low porosity portion remained after the DI measurement. Also, for both cokes, the error bar of the porosity became wider after the DI measurement. This is because cracks have been developed in some coke pieces. The inner crack in the coke A after the DI measurement is shown in Fig. 12. It is possible that the inner pore walls of the coke have been broken. If the DI measurement was continued, the coke would be broken because the crack would grow. Since the high porosity portions disappeared by the volume breakage, the porosity of broken coke piece decreased. While, the porosity of unbroken coke piece was increased by the crack. Because of this error bar of the porosity could become higher after the DI measurement. As shown in Fig. 10, except the results of cracked coke pieces, the average porosity was further decreased and error bar of the both cokes after the DI measurement became small.
Average porosity of cross-section of coke A and B before and after the DI measurement. The right-hand side plots are after the DI measurement without cracked coke pieces.
The high porosity portions (gray color blocks) in the coke A before the DI measurement.
Inner crack in the coke A after the DI measurement.
Pore shape was evaluated by the new parameter Itotal proposed in this work. Itotal of coke before and after the DI measurement is shown in Fig. 7. At first, the effect of HPC on pore shape is described. Pore shape in coke B was much better than that in coke A. Pore shape could have been improved by addition of HPC. It is considered that the coarse pore would be decreased because the coke-matrix would be better connected when HPC was added. Next, the effect of the DI measurement is described. After the DI measurement, Itotal decreased for both cokes. This was because the coarse pores were decreased during the DI measurement. The relationship between I and the number of pores in the unit area of coke A before and after the DI measurement is shown in Fig. 13. Overall the plot of coke pore number after the DI measurement was distributed towards lower side. Also, after the DI measurement, the maximum value of I was decreased and the high I pores, I of which was over 104.7 μm, were disappeared. In case of coke A and B, the high I pores were distributed in the coke inside. The high I pores in the coke A before the DI measurement is shown in Fig. 14. In this figure, black, white and red means pores and background, coke-matrix and the high I pores, respectively. It is difficult to remove the high I pores from this cross-section by only the surface breakage. Therefore, it was considered that the volume breakage occurs at the high I pores.
Relationship between I and the number of pores in the unit area of coke A before and after the DI measurement.
The high I pores (red color) in the coke A before the DI measurement.
The coke-matrix connectivity of coke A and B before and after the DI measurements is shown in Figs. 15(a)–15(d), respectively. In this figure, prime mark (´) means cross-sectional image opposite to the same coke piece. The difference between the two cokes before the DI measurement was the fraction amount of “Over 19 μm connection” and “15.2 μm connection” which had high coke-matrix connectivity. The total fraction amount of these components in coke B was higher than that in coke A. Particularly, the average fraction amount of “Over 19 μm connection” in coke B increased up to 44%, and that in coke A to 27%. It means that by adding HPC coke-matrix connectivity greatly increased. Thus, it was considered that the increase in DI on adding HPC was due to increased coke-matrix connectivity.
Coke-matrix connectivity of coke A and B before (a and b) and after (c and d) the DI measurement.
The surface and volume breakage of the coke due to the DI measurement will remove the defects in the coke. In addition, casting procedures (size and shape adjustment) for embedding coke in resin will lead to “size effect”. The strength of brittle materials such as coke, graphite, and cement etc. increases with decreasing size because of decrease in defects in the materials (size effect).12) Because of all above effects, the observed coke-matrix structure of coke after the DI measurements was more robust than that of before the DI measurement. Comparison of coke-matrix connectivity before and after the DI measurement showed that about “15.2 μm connection” and “19 μm connection” were not destroyed due to the DI measurement. It was estimated that the pore wall with thickness over 15.2 μm was hardly broken by DI measurement. This suggests that cokes with higher sum of “15.2 connection” and “over 19 μm connection” will have higher DI or strength. The sum of “15.2 μm connection” and “over 19 μm connection” was defined as “Connectivity Index (CI)”.
On the other hand, “Non-connected coke-matrix” and “7.6 μm connection” was the major portion of the coke that was removed due to destruction. “Non-connected coke-matrix” was originally had less connectivity and was easy to remove during the DI measurement. “7.6 μm connection” were connected with the thin pore wall which thickness were less than 7.6 μm. These thin pore wall were distributed throughout the cross-section of coke. It was indicated that the thin pore wall could be broken by the impact of the DI measurement. Especially, in case of coke A, a lot of thin pore wall were observed in the coke inside. The thin pore wall distribution in coke A before the DI measurement is shown in Fig. 16. In this figure, gray means the top 10% blocks which have a lot of thin pore wall. These blocks were likely to be broken and would be changed to large pores by the DI measurement. From Fig. 6, it was considered that large pores were easier to have low circularity and, as mentioned above (3.2 Pore structure), these large and coarse pores were cause of the volume breakage. From these results, it was estimated that the thin pore wall was an indirect cause of the volume breakage.
The top 10% blocks (gray color) which have a lot of thin pore wall (less than 7.6 μm) in the coke A before the DI measurement.
Comparison of CI of coke A and B after the DI measurement shows that coke B had higher value than coke A, and that “over 19 μm connection” occupied a large portion of CI of coke B. The reason for this was removal of defects due to the DI measurement and subsequent size effect as described above. Especially the influence of size effect is considered to be more significant. The average size of Coke A was 47.9 mm and Coke B 51.6 mm after the DI measurements. Coke B was more affected by the size and shape adjustment procedure. Because of this, “over 19 μm connection” in coke B became higher than coke A.
Also, the fractures in the cross-sections of coke due to the DI measurement influenced coke-matrix structure. These fractures were observed in the cross-sections of [coke A After-DI-1], [coke A After-DI-1’] and [coke B After-DI-1]. As a result, fraction of the “Non-connected coke-matrix” and “7.6 μm connection” became higher and CI became lower than those of other cross-sections. “Non-connected coke-matrix” in these cross-sections was the broken coke-matrix at the introduction of the fracture. In addition, since the portion with fracture had low coke-matrix connectivity, the fraction of “7.6 μm connection” became higher and CI became lower than those of other cross-sections.
3.4. Cause of Volume BreakagePeirce et al.2) reported that the volume breakage of coke was caused by the following two reasons. One was the cracks that were present inside the coke. The crack was generated due to internal stress from shrinkage during coke-making/carbonization and by the difference in the shrinkage ratio between the coal macerals. The volume breakage caused by such cracks occurred at the beginning of the DI measurement.2) The other reason is fatigue breakage by the impact of the DI measurement.
In the present work, the portion where fatigue breakage may have occurred was clarified by considering the following three portions. First is the high porosity portion as shown in Fig. 10. High porosity will reduce the cross-sectional area for load bearing which will lead to higher stress. When the coke could not withstand the stress, these high porosity portions were destroyed and became the cause of the volume breakage. The second is the portion having coarse pores as shown in Figs. 7 and 13. Basically, pores in coke are formed by escaping gas during coke-making/carbonization. Depending on the timing and/or the amount of gas produced, coarse pores such as large or acute shape were generated in the coke. If it had acute shape, stress would be concentrated to such portions and coke-matrix could be easily broken. The third one is the portion which was not connected to coke-matrix or low connectivity portion. “Non-connected coke-matrix” is considered as cracks in a broad sense. Also, since the thin pore wall (thickness is less than 7.6 μm) is likely to be broken by the impact of the DI measurement, the portion which have a lot of thin pore walls would be changed to a part of coarse pore.
The cause of volume breakage was investigated by image analysis of cross-sectional image of coke before and after the DI measurement. As a result, the following conclusions were obtained.
(1) New image analysis method was developed. This image analysis method evaluated pore shape and coke-matrix connectivity and proposed new parameters “degree of Irregularity (I)” and “Connectivity Index (CI)”. I and CI represents pore shape and coke-matrix connectivity, respectively.
(2) It was revealed that the pore shape and the coke-matrix connectivity were improved on HPC addition.
(3) It was found that volume breakage occurred at the higher porosity portion. As a result, porosity was decreased by the DI measurement. In addition, volume breakage occurred at the portion which had coarse pores. Consequently, the pore shape was improved after the DI measurement.
(4) The volume breakage also occurred at the portion which had cracks or thin pore wall because “Non-connected coke-matrix” and low connectivity portion deceased after the DI measurement. Furthermore, it was clarified that the thickness of the pore wall broken by the DI measurement was less than 7.6 μm.
This study has been carried out as a part of Japanese national project called COURSE50 (CO2 Ultimate Reduction in Steelmaking process by innovative technology for cool Earth 50). We thank NEDO for the support of this study.
