2020 Volume 60 Issue 7 Pages 1545-1550
Softening and melting phenomena of burdens in the cohesive zone in the blast furnace are factors that cause the deterioration of gas permeability. Numerical examinations to predict the softening behavior of burdens in the packed bed were performed in this study. In order to represent the softening characteristics of the simulant sinter ore, the simulation of a single pellet load softening test was carried out. The strength properties of the pellets were evaluated by the shrinkage degree in the load softening test.
Shrinkage degree of pellets with different basicity and pre-reduction ratio were attempted to reproduce. To represent the softening behavior of the pellet, ADEM (Advanced Distinct Element Method) SPH (Smoothed Particle Hydrodynamics) coupling model was applied. Softening and shrinkage could be represented by decreasing the joint spring coefficient while the pellet was loaded. The relationship between the joint spring coefficient and the temperature was determined to compare the shrinkage degree obtained in the experiment and the simulation. The determined joint spring coefficient was used to predict the softening behavior in the load softening test of a packed bed. The simulation results of the shrinkage degree curve showed good agreement with the experimental results. It was indicated that the softening behavior in the cohesive zone could be predicted by this calculation method.
In an ironmaking process, the blast furnace is the multiphase reactor for reducing the iron ore. In this process, iron ore and coke are charged from the top of the furnace, and high temperature reduction gas is introduced from the bottom. As this gas ascends, it reduces and melts the iron ore to form liquid iron and slag in the cohesive zone. The liquid percolates through the coke bed to the hearth. In the reduction process of iron ore, the ore is softened and the ore layer is compressed by the accumulated burdens. It is known that the structural change due to the softening of ore in the cohesive zone has a great influence on the gas permeability in the furnace. The softening behavior of ore is affected by various factors such as chemical compositions, reducing gas composition, temperature, physical properties and so on. To understand the cohesive zone, several experiments have been performed to study the high temperature properties of burdens1,2,3,4,5,6) and the gas flow effect on the liquid flow in the cohesive zone.7,8) Although, it is difficult to say the phenomena in the cohesive zone have been sufficiently clarified yet due to the complexity. Since raw materials for blast furnace vary from year to year, it is necessary to clarify and predict the relationship between the properties of raw materials and the behavior in the cohesive zone.
On the other hand, computer simulation is attracting attention as a technique to mathematically clarify phenomena that are difficult to observe instead of experiment. Several studies that analyzing the behavior of each phase as a continuum has been reported. The analysis of solid, liquid and gas flow in the blast furnace have been performed by using the continuous model.9,10,11,12,13,14) Since discrete solid behavior and continuous liquid behavior coexist in the cohesive zone, it is necessary to treat them simultaneously. It is required to develop a simulation method to predict the behavior of burdens in the cohesive zone from its properties. In previous studies, the authors have developed a coupling model of ADEM (Advanced Distinct Element Method) and SPH (Smoothed Particle Hydrodynamics) and tried to represent the softening and the melting behavior with phase transition.15) The softening and the melting behaviors due to the phase transition are dealt with ADEM.16,17) After phase transition, the melt behavior is dealt with SPH.18,19)
The present work was attempted to predict the structural change of packed bed of iron ore due to softening. The strength properties of simulant sinter ore which is depended on the temperature were represented in the simulation. A numerical experiment of load softening test of a single pellet was carried out. The relationship between the joint spring coefficient which is a parameter representing the strength in the simulation model and the shrinkage degree of a single pellet was obtained. The relationship between the joint spring coefficient and the temperature was determined by the comparison to experimental results of single pellet load-softening test. The softening behavior of the simulant sinter ore packed bed were attempted to predict.
ADEM is a method to extend DEM for analyze the continuum body by considering the interaction force between neighbor connected particles. In ADEM, a cluster particle is represented by agglomerate of primary particles, which are connected each other by joint springs. The joint springs represent the normal and shear interaction forces between two primary particles, respectively. SPH is a lagrangian particle method which does not require a grid and can be used to simulate a compressible fluid moving arbitrarily in three dimensions. Both ADEM and SPH treat the calculation point as particles and calculate the time implementation in an explicit method, so that the coupling of these models are not complicated. Figure 1 shows the calculation flow of ADEM-SPH coupling model. The detailed explanation of this model is indicated in our previous studies.15)
Calculation flow of ADEM-SPH coupling model. (Online version in color.)
In order to represent the softening characteristics of the simulant sinter ore, a numerical experiment of single pellet load softening test was carried out. In this study, experimental results of load softening tests of simulant sinter ore reported by Ohno et al. were used as a reference.20) A softening and melting simulator which can achieve to high speed temperature control over 1000°C/min of sample heating and dry-quenching chamber was used. The load softening test was simulated under the same conditions as the experiment, and the simulation parameters to reproduce the experimental results were searched. The strength properties of the pellets were evaluated by the shrinkage degree in the load softening test. The shrinkage degree of the pellet is calculated from the measured displacement in a vertical direction during the load softening test. The shrinkage degree of the pellet was derived with Eq. (1).
| (1) |
Shrinkage degree of pellets with different basicity and pre-reduction ratio were attempted to reproduce. Experimental methods for sample preparation and load softening test were reported in the literatures.20,21,22,23,24) Sample preparation methods in the experiment were as follows. 3 g of the mixed powder samples were pressed into pellet shape with 15 mm inner diameter die, and they were sintered at 1270°C for 5 min under air atmosphere. The sintered pellet samples, 14 mm diameter and 6 mm height, were treated under reducing gas atmosphere for preparation of pre-reduced samples in 2 levels of different reduction degree. The pre-reduction degree of pellets were prepared to about 30% (low reduction degree) and about 60% (high pre-reduction degree) with CO:CO2:N2 = 1:0:1 at 1000°C for 30 min, respectively. The composition of samples before pre-reduction and pre-reduction degree of each samples were shown Table 1.
| Case | Composition [%] | Basicity CaO/SiO2 | Pre-reduction rate [%] | |||
|---|---|---|---|---|---|---|
| Fe2O3 | CaO | SiO2 | Al2O3 | |||
| #1 | 81.6 | 9.9 | 6.6 | 1.8 | 1.5 | 14.9 |
| #2 | 51.0 | |||||
| #3 | 81.6 | 11.0 | 5.5 | 1.8 | 2.0 | 26.3 |
| #4 | 53.7 | |||||
| #5 | 81.6 | 11.8 | 4.7 | 1.8 | 2.5 | 23.7 |
| #6 | 63.9 | |||||
To confirm the relationship between the joint spring coefficient and the shrinkage degree, numerical experiment of a single pellet load softening test was performed. The joint spring coefficient is as known that it is correlated to Young’s modulus, so that appropriate value to reproduce softening behavior of simulant sinter ore was searched. The load on a pellet was set to 0.1 MPa. The joint spring coefficient was gradually decreased, and the shrinkage degree at that time was obtained to determine the relationship between the joint spring coefficient and the shrinkage degree under the load. Simulation conditions are shown in Table 2. The pellet was represented by an aggregate of a primary particles, and a random arrangement was used for the arrangement of primary particles in the initial state. This is because, as a result of the preliminary examination, the deformation behavior at softening was smoother in the random arrangement than in the regular arrangement (such as simple cubic, body centered cubic, face centered cubic and so on) of primary particles. The primary particle size affects the deformation behavior, it was set to 0.05 mm in consideration of the calculation load. This is because the smaller the primary particle size, the finer the deformation behavior can be expressed, however the calculation load increases. In this study, it was confirmed that the single pellet was smoothly deformed during shrinkage, so that the primary particle size was determined. Figure 2 shows the softening behavior and the relationship between the joint sprig coefficient and the shrinkage degree. Softening starts at 1.4 × 105 N/m of the joint spring coefficient. The shrinkage degree increased with decreasing of joint spring coefficient. As softening progresses, the shrinkage degree increased, and the shape of the pellet was deformed into a flat shape.
| Parameters | Value | |
|---|---|---|
| ADEM parameters | ||
| Primary particle diameter | [mm] | 0.05 |
| Number of particles | – | 9175 |
| Load | [MPa] | 0.1 |
| Joint spring coefficient | [N/m] | 0 − 1.5 × 106 |
| SPH parameters | ||
| Density of fluid | [kg/m3] | 5000 |
| Viscosity of fluid | [Pa·s] | 0.05 |
| Surface tension | [N/m] | 0.47 |
| Contact angle | [θ] | 20 |
Relationship between the joint spring coefficient and the shrinkage degree. The softening behavior of a pellet calculated by the simulation.
In order to predict the softening behavior of burdens in the cohesive zone, it is necessary to clarify the relationship between softening behavior and temperature. By comparing the relationship between the experimentally observed shrinkage degree and the temperature with the simulation results, the joint spring coefficient to represent softening characteristics at a specific temperature were determined. The relationship between the joint spring coefficient and the temperature was determined by combining the shrinkage degree obtained by the experiment20) and the simulation. Figure 3 shows the effect of basicity on relationship between the joint spring coefficient and the temperature of low reduced pellet. Softening of the pellets starts at around 1100°C. The softening of the pellets which basicity 1.5 and 2.0 rapidly progressed at about 1150°C. In the case of the pellet which basicity 2.5, softening was moderate. These results indicate that the higher the basicity, the slower the increase of the shrinkage degree. Note, the simulation does not take into account changes in mineral composition and mineral morphology due to the difference of basicity. The change in the shrinkage degree obtained in the experiment was reproduced by just a change in the joint spring coefficient. The softening behavior depending on the temperature could be represented by the change of the joint spring coefficient depending on the temperature. Figure 4 shows the effect of basicity on relationship between the joint spring coefficient and the temperature of high reduced pellet. Softening of the pellets starts at around 1100°C. The softening of the pellet which basicity 1.5 rapidly progressed at about 1150°C. In the case of the pellet which basicity 2.0, the progress of the softening was slowest. The softening behavior of the pellet which basicity 2.5 was located the intermediate of both pellets. A comparison between the low reduced pellets and the high reduced pellets showed that the low reduced pellets softened at lower temperature. Using shrinkage degree which was obtained experimentally as an index, it was possible to determine the relationship between the temperature and joint spring coefficient of pellet with different basicity and reduction rate.
Effect of the basicity on relationship between the temperature and the joint spring coefficient of low reduced pellet. (Online version in color.)
Effect of the basicity on relationship between the temperature and the joint spring coefficient of high reduced pellet. (Online version in color.)
The load-softening test of a single pellet could determine the joint spring coefficient representing the softening behavior of the pellet. The prediction of softening behavior of the simulant sinter ore packed bed was investigated using the determined parameters. Load-softening test for packed bed was performed using the same apparatus as a single pellet test. The size of the chamber for the packed bed test was 30 mm inner diameter and 40 mm inner height. The pellet preparation conditions were the same as those described above. The pellet basicity 2.0 was used. One pellet was divided into four pieces, and 20 pieces of the 1/4 pellet samples were filled randomly. The pellets were sandwiched by graphite balls due to avoid contact between the pellet and crucible. The filling height of the entire packed bed including the graphite balls was 30 mm, and the height of pellet layer was 12 mm. The schematic diagram of a packed bed type sample in sample holder is shown in Fig. 5. The samples were heated up to 900°C with 1000°C/min. After reach to 900°C, a heating rate was changed to 10°C/min. During heating experiment, 0.1 MPa load was added to the sample from top art through lid and N2 gas flowed from bottom side with 1 NL/min. Displacement amount of the sample thickness was measured for evaluation of deformation behavior. The simulation was performed under the same conditions as the experiment, however the following points differ from the experimental conditions. The graphite balls above and below the packed bed were ignored and only pellets were calculated. In order to decrease the calculation load, the temperature was raised from 900°C to 1500°C in 10 seconds. This condition was same as 3600°C/min heating rate. In the experiment, the generated melt flowed out of the crucible through the graphite ball layer, however in the simulation, the melt did not flow out.
Schematic diagram of a packed bed type sample setup in sample holder.
Figure 6 shows softening behavior of simulant sinter ore packed bed in simulation. Shrinkage of the packed bed was observed with increasing of temperature. Figure 7 shows the shrinkage degree with temperature both experimental and simulation results. The shrinkage degree rapidly increased at about 1200°C and shrinkage progressed to about 1300°C and then converged. The simulation results showed same trend in same temperature range. The shrinkage degree of the low reduced pellet higher than the high reduced pellet in the experiment. On the other hand, the difference of shrinkage degree curve between low and high reduced pellet in the simulation was relatively smaller than the experimental one. The reason of this is considered that in the load softening test of a single pellet, the pellet received uniaxial direction load. However, the pellet received the load in the multiaxial direction in the load softening test of the packed bed. Since the shrinkage degree measures only the deformation behavior in the uniaxial direction, a slight difference from the shrinkage behavior in the multiaxial compression due to the pre-reduction ratio may have occurred in the pellet. It may also be influenced by the heating ratio in the simulation that is much higher than that of the experiment. Although the difference in the shrinkage curve due to the difference in reduction ratio was smaller than that in the experiment, the softening behavior of the packed bed was reproduced approximately. These results indicate that this simulation method can be a useful tool to predict the softening behavior in the cohesive zone. In the future works, prediction of detailed change in the structure of packed bed and accompanying changes in pressure drop will be analyzed.
Softening behavior of simulant sinter ore packed bed in the simulation.
Effect of reduction rate on relationship between the temperature and the shrinkage degree. Comparison of experimental22) and simulation results. (Online version in color.)
To predict the softening behavior of the simulant sinter ore packed bed, numerical analysis was performed using ADEM-SPH coupling model. The softening behavior of a single pellet was represented by the simulation. The behavior was evaluated by shrinkage degree. The following results were obtained.
(1) In the simulation, the softening behavior of the pellet could be expressed by decreasing the joint spring coefficient. The relationship between the joint spring coefficient and the temperature was determined using the shrinkage degree obtained by the load softening test of a single pellet as an index.
(2) The determined simulation parameters were used to predict the softening behavior in the load softening test of a simulant sinter ore packed bed. The simulation results of the shrinkage degree curve showed good agreement with the experimental results.
This work was supported by the JSPS KAKENHI Grant number 19K15331, 15H04168 and the research group of Control of Cohesive Phenomena in Blast Furnace to optimize Gas Permeability established in Iron and Steel Institute of Japan.
