Understanding Ironmaking in Blast Furnace and Dissection studies in Japan
Understanding Ironmaking in Blast Furnace and Dissection studies in Japan
The blast furnace (BF) ironmaking is the most viable means of producing hot metal (HM) mainly due to its well established and proven performance, flexible raw material usage, and high thermal energy conservation capability. There are no definitive dates available for the inception of BF ironmaking. However, important process designs and re-engineering began to be implemented in the ironmaking furnaces in Europe as far back as the 14th century. Since then, the BF route has dominated as a process of preference over other alternative iron production methods.
Since its inception, BF ironmaking process, in order to sustain and remain viable, has undergone enduring evolutionary developments to become a highly efficient process. Most important developments to date include (i) facility modernization, (ii) increased furnace productivity, (iii) reduced coke rate, (iv) extended furnace campaign life, and (v) materials flexibility and improvements. The technological advances which have been carried out to improve the economy, efficiency, and making the process environmentally friendly include (i) various process management and control practices, (ii) centre coke charging, (iii) high furnace top pressure operations, (iv) furnace operation with oxygen (O2) enrichment, (v) substitution of expensive coke with alternative supplementary carbon sources namely pulverized coal, natural gas, oils, and plastics, and (vi) many more. The technological development of the BF ironmaking process has led it from a small production unit, consuming large quantities of fuel to its present state where furnaces designed to produce 10,000 tons of HM per day are fairly common in several countries. Even with this dramatic increase in size and output of the BF, many of the reactions occurring within the furnace are still not known.
Modern BFs operate predominantly on sinter, pellets, sinter/sized iron ore, or sinter/pellet burdens. The type and quality of these materials depend upon individual plants operating philosophy, thus the production and reduction characteristics of these materials are of supreme importance to the BF process.
The development of the BF process
The developments which have taken place in the BF process have been introduced gradually, finally becoming standard operating practices. It is possible to list the major developments in an approximate chronological order consisting of (i) effective use of prepared burdens, (ii) blast injectants, (ii) high top pressure, (iv) high blast temperature, and (v) improved control of burden distribution. Simultaneous to these developments, there has been gradual increase in physical size of the BF.
Use of prepared burdens
The first preparation of burden materials was merely the sizing of the iron ores charged to the furnace. The closer sizing of the burden improved the permeability of the furnace, allowing more wind to be blown, thus increasing the output of the BF. Additionally the efficiency of the reduction reactions also increased because of the removal of the larger ore lumps, thus decreasing the coke rate.
Sintering of iron ores prior to charging was the second important step in burden preparation, but sintering was originally developed to render such waste iron bearing materials as BF flue dust, mill scale and ore fines into a usable BF feed. However, this concept changed rapidly after the success achieved with the self-fluxing sinter in the mid 1950s, allowing the fluxes to be removed from the BF charge burden and to be introduced through the sinter. This has also led to the reduction of the coke rate while increasing the BF productivity. The present situation is that sinter is now an established burden component in modern BF plants and it is still under continuous investigation in order to further improve its properties.
The depletion of readily available high grade iron ores made it necessary for suppliers to upgrade their product by means of beneficiation followed by the production of high grade iron ore pellets from the concentrate. This process gained wide acceptance resulting in the burden charged to the BFs furnaces having high iron content and a low gangue content, which in turn led to a further increase in the production coupled with a fall in coke rate. Another improvement claimed with pellets has been that the permeability of the burden increased due to the closer sizing. However, the use of pellets is not universal and in several countries BFs are being operated predominantly on sinter burden with pellets and /or sized iron ore contributing a small proportion of the overall charge. Indeed there are some views that a high sinter ratio is essential for the stable operation of large BFs because sinter has superior high temperature and better distribution properties, since pellets roll easily, making control of distribution difficult. Also due to the use of sinter in the BF, the coke breeze generated at the coke oven plant gets consumed within the steel plant.
There are three injectants which are normally being used in the BF. These are (i) steam, (ii) O2, and (iii) auxiliary fuels. Injectants affect the flame temperature, i.e. the temperature of the flame in the combustion zone of the tuyere. Steam and auxiliary fuel lower the flame temperature whilst O2 increases it. The theoretical flame temperature can be calculated and it is extremely important for maintaining smooth operation and raising productivity with large quantities of O2 and auxiliary fuel. A low flame temperature prevents reaction in the furnace and leads to furnace chilling. A high flame temperature can damage the permeability as a result of expansion of the melting zone and evaporation of alkalis and silica (SiO2) in the burden. In order to ensure a smooth operation, it is necessary to control the theoretical flame temperature by varying the quantities of the injectants.
Looking at the three injectants separately, steam reacts with coke to produce hydrogen (H2) which increases the extent of reduction of burden materials, thus decreasing the fuel rate. Auxiliary fuels are the main injectants. The type of auxiliary fuel used depends upon local conditions. Injection of auxiliary fuel provides additional quantities of H2 and carbon monoxide (CO) in the BF, increasing the degree of reduction of the burden, which in turn provides for a decrease in the coke rate. It is essential to ensure complete combustion of auxiliary fuel at the tuyere since incomplete combustion can impair the furnace permeability, causing adverse effects upon furnace operation. Insufficient combustion can be controlled by providing sufficient excess O2 in the blast.
Oxygen injection increases the quantity of excess O2 and increases the flame temperature, which counteracts the drop in flame temperature caused by the injection of steam and auxiliary fuel. It is also useful for decreasing the quantity of bosh gas, thus minimizing the extent of channeling of the gas in the BF and the extent of flooding and loading. The flooding causes irregular furnace operation. The loading is the situation when the molten slag is prevented from descending due to the upward gas velocity. Ultimately the weight of the slag is to become sufficient to overcome the gas flow for it to descend. When the ratio of O2 in the blast is gradually increased, the volume of gas produced per ton of HM decreases, causing a drop in the quantity of heat transfer from the gases to the solids in the shaft, resulting into a decrease in the temperature of the shaft. Also, the increase in productivity through O2 enrichment accelerates the burden descent rate, resulting in less time for heat transfer. As a result, the burden materials enter the high temperature zone without being sufficiently pre-heated, thus chilling the furnace and causing slipping and hanging of the burden.
The practice of using injectants requires careful control with regards to three limits namely (i) limit in heat transfer between gases and solids, (ii) limit of theoretical flame temperature, and (iii) limit of complete combustion of the auxiliary fuel. Control within these limits provides low fuel rates and high productivity.
High top pressure
The benefit of the high top pressure is that it reduces the gas velocity in the furnace thus allowing more time for gaseous reduction resulting into decrease in the fuel rate. Dust losses also decrease because the lower gas velocity is insufficient to convey the coarser dust particles. Alternatively, more wind can be blown and thus production can be raised while maintaining the same gas velocity in the furnace, hence preventing channeling, flooding, and loading. The main disadvantage is that to accommodate the increased gas pressure, sturdy construction is needed throughout the BF equipments from stoves, through bustle pipe, furnace walls, furnace top, and gas cleaning plant etc., which is obviously expensive. Certainly the BF top alone needs special design in order to equalize gas pressure in the charging system and prevent wear of the furnace top charging equipment. Another disadvantage is the loss in energy in the high pressure top gas, although the top gas recovery turbine can recover part of this energy.
Although there are issues with the application of high top pressure from an engineering aspect, the operation of large blast furnaces requires the use of it to (i) lower the fuel rate, and (ii) permit higher productivity of the furnace.
High blast temperature
The air entering the BF through the tuyeres is heated by the combustion of the coke, and hence, the hotter the incoming air, the lesser is the coke consumed in further heating inside the tuyere region. Pre-heating of the air is not a new thing. Indeed over a century ago the BF stoves were in existence. However, it is only relatively recently that temperatures in excess of 1300 deg C have been achieved. The achievement of higher temperatures is due to modifications to stove design. These modifications are (i) increasing the area of heated surface of chequer work by alteration of the shape of the bricks, (ii) using a higher quality refractories which are able to withstand higher temperatures, and (iii) providing external combustion chambers which also increases the heated surface area.
Improvements in the burden distribution
Control of the distribution of burden materials is important for improving the gas utilization and lowering the fuel rate. Correct distribution is also necessary to control the shape of the cohesive zone of the BF, thus maximizing the production and minimizing the gas flow at the BF wall, the latter prolonging the life of the furnace.
As furnace size increased, the distribution of burden material required to provide a stable gas distribution inside the BF cannot be maintained with conventional charging equipment due to differences in compiling angle, density and shape. These problems have been partially overcome by the installation of movable armour to control the distribution of material as it leaves the charging bell. The development of the bell-less top and Gimble top charging systems, which are having a rotating chute capable of accurately distributing the charge to any position on the furnace stock-line, have to a great extent helped in solving the issue.
Iron ore sinter
In many BFs, sinter is the major component of the BF burdens. The chemical composition of the sinter depends upon the other components constituting the furnace burden. Normally the sinter ranges from fluxed (CaO/SiO2 around 1.2) to super-fluxed (CaO/SiO2 around 1.7 to 2.2). The fluxed sinter is generally used when the majority of the furnace burden is sinter. Super-fluxed sinter is used when the rest of the burden is acidic in nature, thus balancing the slag chemistry to provide an acceptable slag composition. Sinter is extremely heterogeneous due to the nature of the sintering process.
Structure of the sinter – The fluxes, which are mixed with the iron ore, react during sintering, melt and attack the ore particles. Complete agglomeration of small ore particles can occur, but generally the larger particles only undergo surface attack. During cooling, precipitation of different phases occurs within the slag matrix, the overall result being a mixture of phases which are far removed from an equilibrium state and a heterogeneous material whose phases depend upon the segregation of components in the initial mixing, e.g. where particles of lime existed prior to sintering, a lime rich region is formed. Overall the phases present depend upon the quantity of fluxing agents added. Self-fluxing sinter is mainly hematite and magnetite with small quantities of calcium ferrites, produced by a reaction between iron ore and lime. Here the term ‘ferrites’ refers to the combined quantities of the different species of ferrites which can be produced, depending upon the basicity and the ore particles in the reaction zone. As the basicity increases, there is an increase in the proportion of ferrites.
In sinter, normally, the ferrites are contaminated with SiO2 and Al2O3 (alumina) and the product is known as SFCA (silico-ferrite of calcium and alumina). The SFCA normally conformed to a general formula ‘n1(Fe2O3).n2(SiO2).n3(Al2O3).5CaO’, where the sum of n1,n2,and n3 is around 12. The calcium content is fairly constant at around 15 %. In practice usually in sinters the ferrites are usually found are 7 Fe2O3.2SiO2.3AI2O3.5CaO, and 9Fe2O3.2SiO2.0.5AI2O3.5CaO.
Reduction of sinter – The type and quantity of ferrites present in sinter play an important role in the reduction properties. The reducibilities of the ferrites are not constant, but vary from one kind to another. It is seen that the proportion of ferrites increases as the sinter basicity rises. However, the reducibility does not follow the same trend. Between the basicity ranges of 1.0 to 1.5, the reducibility rises due to the increase in ferrites of the type CaO.2Fe2O3 and CaO.FeO.Fe203. At a basicity range of 1.4 to 1.5 the reducibility decreases due to a drop in the proportion of hematite present in the sinters and the disappearance of CaO.2Fe203 coupled with the appearance of the relatively not reducible 2CaO.Fe2O3. The increase in basicity beyond 1.5 again shows an upward trend due to the appearance of CaO.Fe2O3 and CaO.FeO.Fe2O3.
The reduction behaviour of ferrites is complex in that they are to decompose for reduction of the iron oxide to occur. During the reduction process, first the higher iron oxides and ferrites, rich in iron oxide, are reduced until only di-calcium ferrite and wustite remain. The gas then attacks the di-calcium ferrite as per the reversible reaction 2CaO.Fe2O3 + 3H2 = 2CaO + 2Fe + 3H2O. The liberated CaO then immediately reacts with the wustite as per the reversible reaction 2CaO + 3FeO = 2CaO.Fe2O3 + Fe. The reaction then proceeds as per earlier equation and so on. However, micro-photographs show that wustite is not present at the gas boundary and hence a diffusion process between the two reactions is to take place. Studies have indicated that at the oxide surface, the di-calcium ferrite is first reduced. The liberated iron separates out in the oxide phase and the calcium diffuses in and reacts with the wustite and again the iron either separates out or diffuses into the Fe3O4.
Iron ore pellets
During the production process of iron ore pellets, the iron ore is beneficiated by crushing and removing the liberated gangue material. Generally some quartz is added in the production of acid pellets for improving the pellet properties. The majority of pellets produced are of the acid type, i.e. without any intentional substantial addition of the flux. During the production of acid pellets, green pellets are fired at around 1300 deg C in an oxidizing atmosphere. This promotes bonding of the particles by (i) sintering of the hematite grains, (ii) oxidation and subsequent sintering of the magnetite grains, and (iii) slag bonding. The latter is caused by fusion of the small traces of gangue and the bentonite, used in the process of pelletizing to ensure sufficient strength of green pellets. This slag phase consists essentially of lime, silica, ferric oxide and small traces of alkalis, magnesia, alumina, etc.
An indication of the chemical composition of the slag phase can be obtained by referring to the CaO-SiO2-Fe2O3 phase diagram. A point to be noted is that equilibrium diagrams are to be used carefully as in the majority of processes the reactions are rarely at equilibrium, nevertheless, such diagrams are useful tools. Very little reaction, if any, occurs between the quartz grains and the hematite during firing and thus acid pellets are comprised of hematite, quartz, a slag phase, and in some cases, if sufficient firing has not taken place, magnetite, originating from any magnetite ores in the pellet blend.
The acid pellets are used in some BFs in the furnace burden. The amount used in the burden depends upon the operating practice adopted. In case of BFs operating entirely on acid pellets as the source of iron units, the flux (limestone and dolomite) needed for the slag formation process is charged in the BF as part of the burden.
Fluxed pellets – Presently the use of fluxed pellets is preferred. In the fluxed pellets, the fluxes are incorporated into the pellet, hence avoiding the necessity of charging them separately into the furnace. Fluxed pellets can be produced with lime additions, as the flux, or dolomite. As the basicity of the pellets increases with the addition of the flux, there is a change in the microstructure takes place. Considering lime fluxed pellets, the addition of lime has an influence on the slag composition and quantity, also the quantity of hematite. The addition of lime creates the possibility of a reaction between the hematite and the lime to produce calcium ferrites CaO.Fe2O3 or 2CaO.Fe2O3 depending upon the lime concentration. With fluxed pellets, the firing temperature is lower than that of acid pellets to avoid the formation of excessive slag.
In case of fluxed pellets, it is expected to find grains of hematite surrounded by calcium ferrites caused by the chemical reaction by lime. In some cases, the original hematite grain can be fully converted to calcium ferrites, which clearly depends on the original hematite grain size. The effect of lime on the slag phase is two-fold. Firstly there is a general increase in quantity of the slag and secondly a change in basicity. The exact composition naturally depends upon the quantity of phases reacting, but the possibilities can be surmised from the CaO-Fe2O3-SiO2 phase diagram. One of the issues with the fluxed pellets is their relatively poor reduction properties. This shortcoming of lime-fluxed pellets has led to the production of pellets fluxed with dolomite, instead of lime.
The addition of magnesia to iron oxide results in a solid state reaction between the two and an increase in melting temperature. Hence in dolomite fluxed pellets magnesio-ferrites MgO.Fe2O3 or (Mg.Fe)O.Fe2O3 are produced. The quartz cannot be fully absorbed in dolomite fluxed pellets because melting between magnesia and silica does not occur at the firing temperature and only reactions which occur in solid state can take place.
Reduction mechanisms associated with acid pellets can be explained by gaseous reduction, reaction kinetics, and direct reduction. In case of gaseous reduction, as the O2 is removed from the iron oxide, acid pellets follow a reduction path of hematite to magnetite to wustite (at temperatures higher than 560 deg C) to metallic iron. These phase changes are represented by the reversible gaseous reactions, using CO as the reducing agent. The equations are 3Fe2O3 + CO = 2Fe3O4 + CO2, Fe3O4 + CO = 3FeO + CO2, Fe3O4 + 4CO = 3Fe + 4CO2, and FeO + CO = Fe + CO2.
The mechanism of hematite reduction has been extensively studied and it has been noted the reduction of hematite does not take place in discrete steps, i.e. to magnetite, then to wustite, etc., but reduction produces a top to bottom chemical structure, provided the reduction potential of the gas is high enough, i.e. the structure consists of a hematite particle, surrounded by a layer of magnetite, then wustite and finally an outer layer of metallic iron. Wustite is non-stoichiometric, i.e. it is deficient in iron ions. These vacancies are the important defects in the reduction behaviour of iron oxides as they make possible the diffusion of iron through the iron oxide lattice. The removal of O2 from wustite produces a filling up of the iron ion vacancies at the oxide surface.
The surface reduction sets in motion a diffusion of vacancies and electron defects from the interior of the oxide towards the reaction interface. With the reduction of wustite the inward flow of metallic ions react with the magnetite layer, thus reducing the magnetite. The reaction then takes place and the cycle repeats itself gradually reducing the magnetite.
The kinetics of the reduction of iron oxides has been extensively studied bur there exists some conflicting views as regards to the rate controlling step. The process of gaseous reduction of iron oxides necessitates many steps such as (i) diffusion of the reacting gas from the bulk gas phase through the boundary layer, (ii) diffusion of the gas through the product layer to the reaction interface, (iii) adsorption of the gas onto the reaction interface, (iv) chemical reaction at the interface, (v) desorption of the product gas from the reaction interface, (vi) diffusion of the gaseous reaction products away from the reaction interface to the particle surface, and (vii) diffusion of the product gas through the boundary layer into the bulk gas phase.
Although great conflict exists as to the rate limiting step or steps, generally reduction of iron oxides conform to the equation derived by McKewan K1 = Kw/do = ro[1 – (1- R1/3)]/t where K1 is the rate of advance of the hematite/magnetite interface in mm/minute, Kw is the rate constant in g / sq mm / minute, do = density of the pure iron oxide sphere in g/cu mm, ro is the radius of iron oxide sphere in mm, R is the fractional conversion of hematite to magnetite, and t is the reaction time in minutes. It is claimed that as the reduction rate of iron oxides conform to this equation, the rate limiting step is the chemical reaction.
Hills used mass transport principles to show that a reaction controlled by mass transfer and diffusion alone can have the specific characteristics frequently used to identify a chemically controlled reaction, particularly the linearity of [1 – (1 – R)1/3)] with time. Hills postulated that the reaction is controlled by both processes of (i) gas diffusion through the product layer and (ii) transport through the boundary layer external to the particle. A form of Hills rate equation can be expressed as 3[1 – (1 – R)2/3]- 2R(1- Bm) = C2.t where R is the fractional reduction, t is the reduction time in seconds, Bm = DE/Kg.ro, Bm is the modulus for mass transfer, i.e. the ratio of the diffusion resistance within the product layer and mass transfer resistance outside the particle, DE is the diffusion coefficient in the product layer in sq mm/sec, Kg is the mass transfer coefficient to the surface of the reacting sphere in mm/sec and ro is the radius of sphere in mm. C2 is a constant for a reduction reaction and depends upon the iron oxide sphere properties and the environmental conditions.
In the case of reduction at temperatures of 800 deg C and higher, an increase in temperature of gaseous reduction leads to a rise in the reaction rate, provided melting of the particles does not occur. An increase in porosity also produces a rise in the reduction rate.
The mechanism of direct reduction of iron oxides with carbon (C) is extremely important in the BF and it has been found that direct reduction only occurs in appreciable quantities at temperatures in excess of 900 deg C. The direct reduction reaction can actually be split into equations FexOy + C = FexO(y-1) + CO. The reduction by gas is CO + FexOy = FexO(y-1) + CO2. In these reactions x = 1, 2 or 3 and y = 1, 3 or 4. The solution loss (Boudouard) reaction CO2 + C = 2CO provides CO for the gaseous reaction. Since, the direct reduction reaction actually occurs via an indirect reduction reaction, indicates that the direct reduction of solid oxides in the BF process is of no importance with respect to the progress of the reaction.
The situation with liquid iron oxides reacting with C is of course quite different in that the area of contact between liquid oxide and solid C is much greater than that of solid oxide and solid C. Also, diffusion of reacting and product species in the liquid is much faster than that in the solid state. These effects lead to much higher reaction rates in liquid oxide / solid C systems than in solid oxide / solid C systems. The higher is the reaction temperature the greater is the degree of the reduction. It has been found that the rate limiting step for reduction of FeO rich slags changes with the extent of reduction. Nucleation of reduced iron and chemical reaction at the C / liquid interface appear to constitute the rate limiting step until a high degree of reduction is attained. At higher reduction levels the slowest step is the diffusion of O2 through the slag boundary layer.
Influence of alkali upon reduction
Alkali re-circulates inside the BF by vapourization in the high temperature zone and subsequent deposition on the burden and coke in the cooler regions. The deposited alkali then descends with the burden and coke to be finally vapourized. The nature of this re-circulating effect is such that quite high levels of alkali can accumulate inside the BF which can influence the reduction of the burden materials. Alkali additions are found to increase the reduction rate of acid and basic pellets. It has been noticed that there is an optimum level of alkali additions, above which the reduction rate is decreased due to the extensive slag formation. Also, dolomite fluxed pellets show a decrease in reduction rate when there are alkali vapours in the reducing gas. Further, the type of alkali type is important, i.e. sodium hydroxide (NaOH) is better promoter of the reduction reaction than the sodium chloride (NaCl) for the same concentration.of the sodium cation.
The increase in reduction rate which occurs when alkali is added to the iron oxide is caused by the increased swelling exposing a greater surface area to the reducing gas. In addition the alkali causes non surface related chemical reduction which means that the surface of the wustite is continually exposed to the reducing gas instead of being shielded by a layer of metallic iron. Non surface related chemical reduction is caused by the incorporation of alkali cations into the wustite lattice which causes in-homogenization in the wustite activity, modifying the nucleation behaviour of iron, thus leading to non surface related chemical reduction. Pellet swelling seems generally regarded as a symptom of excessive alkali content.
Burden behaviour in the BF
In the lifespan of the BF process a considerable wealth of knowledge of the reduction characteristics of iron oxides, pellets and sinters has been accumulated, upto reaction temperatures of around 1000 deg C. At temperatures beyond this, very little is known of the reactions occurring or their effect on the properties of BF burden materials. Even with the vast quantity of information available on burden material behaviour at temperatures upto 1000 deg C, it is not easy to apply it for the simple reason that internal examination of the BF during operation is extremely difficult. The main ‘tools’ for obtaining samples from operating BFs are gas probes, temperature probes, and burden probes etc. although their useful coverage is only a very small volume of the BF.
It is fair to state, however, that correlations between material behaviour during reduction and BF process have been reasonably established. As an example, it is known that materials which show a large degree of breakdown of physical size during low temperature reduction cause a decrease in furnace permeability in practice and highly reducible burden materials decrease the fuel rate. Also pellets which swell extensively during reduction lead to a loss of furnace permeability.
A major breakthrough concerning the behaviour of materials within the BF came with the water quenching of several operating furnaces and the methodical dissection and study of their contents. The distribution of the burden inside the BF depends upon the charging sequence, charge weights, burden components, and furnace operation and results in each furnace operating in a different manner. Fig 1 shows the internal structure in Kukioka BF no.4. The ore and the coke layers are maintained until the softening-melting zone or cohesive zone is reached. The start of the cohesive zone during the dissection procedure has been ascertained by the increased physical resistance to removal of the material by the mechanical means. The cohesive zone is where the materials start to soften and eventually melt. The discovery that the cohesive zone is not in one region of the furnace, but is distributed in a reasonably geometrical shape was one of the major insights to the reactions occurring within the BF during its operation.
It was found that the structure of the cohesive zone varied depending upon the furnace operation. As an example, Fig 1 also shows the structures found in three different furnaces. Hirohata BF no. 1 shows the softened layers having a ’doughnut’ shape arranged in an inverted ’V’ structure, whilst Kukioka BF no.4 has a ’W’ shaped cohesive zone. Higashida BF no.5 reveals a distorted inverted ’V’ caused by irregular furnace operation prior to the quenching operation.
Fig 1 Dissection studies of Japanese blast furnaces
Reduction levels – The study of the extent of reduction in each burden layer of Hirohata BF no. 1 and Kukioka BF no.4 has brought out interesting features. One of the interesting features is the fact that very little reduction occurs until the burden reaches the cohesive zone, wherein reduction proceeds rapidly. One of the major problems with water quenching is the possible reoxidation of the burden material during the cooling period and laboratory tests were conducted to determine the extent of reoxidation which might be taking place. One study was made to measure the reoxidation of sinter, in the laboratory, under the same cooling conditions existing during quenching of a BF, using a series of different initial reduction levels. The another study used another technique employing burden materials of various reduction degrees cooled from three different temperatures (400 deg C, 800 deg C and 1000 deg C) at a cooling rate of 200 deg C per hour in a nitrogen (N2) atmosphere. In this study it has been found that although the reduction temperatures and reduction degrees were different, the final reoxidation degree was around constant at 20 % to 25 %, i.e. the reoxidation increased in proportion to the initial reduction degree. At temperatures below 300 deg C, no reoxidation occurred. The result of these experiments is that the reduction levels were required to be increased, for example, from 10 % to 30 % to 15 % to 40 %. These corrected levels were in agreement with the reduction levels found in Russian dissection studies on a N2 quenched furnace.
Temperature profiles – The temperature isotherms within the furnaces were estimated by a combination of several methods. In one method, ‘Tempil’ pellets encased in numerous graphite holders were charged prior to blowing out the furnaces. This technique allowed the estimation of the temperature within the range 200 deg C to 1800 deg C, but one of the problems with this technique was that there was no method of controlling the distribution of the graphite holders within the BF. The other methods employed were measurement of the extent of coke graphitization, thus estimating the temperature between 1200 deg C and 1700 deg C. Measurement of the coke electrical resistance, which allowed temperature estimation between 1100 deg C and 1700 deg C and finally the degree of iron ore fusion was measured to estimate temperatures within the range of 900 deg C to 1400 deg C.
Comparing the isotherms with the distribution of the softening-melting burden layers (Fig 1 and Fig 2), it was found that the cohesive zone exists over a temperature range of around 1100 deg C to 1500 deg C for BFs operating mainly on sinter burdens.
Fig 2 Estimated temperature profiles in Kukioka BF no. 4 and Hirohata BF no. 1 and structural of the softening-melting layers in Hirohata BF no. 1.
Burden layer structure within the cohesive zone – The type of structure of an individual burden layer in the cohesive zone depends upon the position of the layer within the BF. Two layers from Hirohato BF no.1 are shown in Fig 2. Layer G-5 is near the apex of the cohesive zone, while layer G-19 is situated near the base of the cohesive zone. Layer G-5 has four distinct zones, two of which are lumpy or granular portions (C and D). Layer G-19, on the other hand, contains only one lumpy portion, D. Apart from the obvious shape differences between the layers, the other main difference is the replacement of the icicles’ in layer G-5 by a half-molten portion just prior to melting down, A in layer G-19.
As seen earlier a substantial amount of reduction takes place in the cohesive zone and this has been proved by the reduction data obtained for each portion as given in Tab 1, and Tab 2. The figures are on the low side, as reoxidation, caused by the act of water quenching, certainly have taken place. The reason for the high reduction level of portion C is attributed to the slightly lower reduction temperature while in contact with the coke.
|Tab 1 Degree of reduction of the burden materials in the softening-melting layers of Hirohata BF no. 1|
|Softening-melting layer||Portion*||Reduction degree %|
|* B: Softening portion, C and D: Lumpy portion|
|Tab 2 Degree of reduction of the pellets in the lumpy portion|
|Softening-melting layer||Sampling position (distance from the boundary*) (m)||Mean value of the reduction of the sample pellets (%)|
|* Between the lumpy and softening portions|
|** The value of the reduced pellet being not reoxidized|
The thickness of the softening-melting layers in Hirohata BF no. 1 ranged from 400 mm to 500 mm, in the case of the upper layers, to 70 mm – 100 mm for the layers near the base of the cohesive zone. The diminishing thickness is due to compaction, caused by the pressure exerted by the weight of material above the layer and also because of a natural thinning of material due to the increase in furnace diameter as the material descends. In the softening portions iron ore granules were combined in contact with each other. Sinter particles in the layers deformed very little, unlike pellets, which showed signs of deformation.
The process of pellet metallization can take place in one of three modes namely (i) the metallic iron is uniformly distributed within a pellet, (ii) a metallic shell is formed, leaving a wustite core, and (iii) wustite within the pellet reacts to form a slag and moves towards the metallic iron shell, leaving a central cavity. The reason for these three possible modes is not connected with the distribution within the softening-melting layer, but can be due to differences between the pellets or uneven gas flow in the softening-melting layer.
It has been found that the half molten portion consisted of highly compacted metallic iron and a small quantity of slag. Any limestone or olivine present remained unslagged. The icicles extend into the coke voids and consist of a metallic shell with a hollow interior, with small droplets of slag adhering to the iron. The higher the softening-melting layer within the furnace, the greater the length of the icicles, e.g. level G-1 produced some icicles of several hundreds of millimeters in length, while the lower layers produced icicles only 10 to 20 millimeters long.
The structure of the softening-melting layers in Kukioka BF no.4 was basically identical to those described for Hirohata BF no.1, except the thinner burden layers made the structure less distinct and the icicles smaller.
Slag composition changes – The major chemical change of the slag phase in the softening-melting layers is a decrease in the FeO content as the slag trickles down from the melting portion. Although large differences were detected by x-ray microanalyses of slags in portion A, ranging from 2 % to 20 % FeO, depending upon the location, the FeO content of the slag immediately prior to separation from the softening-melting layer was only 2 % to 3 %. The type of slag was not significantly different to that found in the normal sinter product, but in the ore granules a considerable quantity of fayalite was produced. Descent of the slag results in a gradual change in composition. The gradual increase in the CaO/SiO2 ratio is attributed to fluxing with limestone and a drop in the SiO2 content, caused by SiO2 reduction. The rise in Al2O3 is created by the incorporation of coke ash into the descending slag.
Metal composition changes – Considering the changes in metal composition as it descends the furnace; the carbon content of the metal in the half-molten portion of the softening-melting layer is around 0.2 % in the upper part and 0.35 % to 0.57 % in the lower part. The source of C in these half-molten layers is attributed to the carburizing action of the CO, except for the metal in contact with coke. Similar trends are visible in the layers found in Kukioka BF no.4. The rise in the C content of the icicles is attributed to the metal being in direct contact with particles of coke. Two distinct processes have been identified which are operating for the separation of metallic iron from the layers. The first mechanism is via the icicles which form at 1350 deg C to 1400 deg C and drip into the coke bed. Reduction of the iron oxides present in the icicles occurs rapidly to produce metallic iron. The second process occurs in layers in which no icicles form. In this situation, the metallic iron is carburized by the underlying coke until it reaches a C level such that melting can occur at the pertaining temperature. In this case the temperature of meltdown is around 1500 deg C.
The question of the mechanism of silicon pick-up by the metal within the furnace has been the subject of considerable discussion. Studies carried out in the experimental BF at Liege, Belgium fitted with sampling probes have found that the silicon level rise gradually from the melting zone to the hearth, such that 75 % of the final HM silicon is achieved by the time the metal reached the tuyere level. The Japanese dissection studies on the other hand reveal that the silicon level of the metal at the tuyere level is far in excess of that of the tapped HM. An explanation for this discrepancy between the two groups of studies can be that silicon pick-up had occurred during the process of water quenching the Japanese furnaces. During the experiments conducted to determine the probability of silicon pick-up during quenching, it was found that silicon pick-up from any slag present could be a possibility. Hence, this is to be borne in mind when analyzing the Japanese dissection data.
The sulphur (S) level of the metal within the softening-melting region is much higher than the concentration in the tapped HM. In the granular zones very little increase in S level occurs, which can be due to the materials in the softening-melting zone absorbing the S from the ascending gases, rather than a lack of absorption capacity by the burden in the granular zones. The lack of substantial quantities of S in the gas in the stack of the furnace can explain the horizontal profile at temperatures below 800 deg C. Further, as the temperature and slag basicity rise, the distribution of S between the slag and metal increases accordingly. Some idea of how S recirculates within the BF can be seen in Fig 3 in which the circulation of S within Hirohata BF no. 1 is shown.
Fig 3 Circulation of sulphur within Hirohata BF no. 1
Size distribution – The change in physical size of the burden components during their descent was determined from the quenched furnace data and one of the major problems with this part of the study was that breakdown of material occurs during the quenching operation. Degradation of sinter reaches a maximum at temperatures of 400 deg C to 600 deg C and increases with the retention time. At levels of reduction in excess of 30 %, very little degradation occurs. Estimation of the cooling pattern of Kokura BF no.2 shows that the burden materials are exposed for a lengthy period of time to conditions which lead to considerable breakdown. The effect of the water quenching operation on the degradation of sinter was calculated. This calculation indicates that the sinter degradation increases with time after blow out and considerable degradation occurs in the region around the middle of the shaft.
Applying this to a centre working furnace (centre working means that the majority of the gas flows up the central axis of the furnace), it has been noticed that the degradation of sinter in the central zone of the furnace, where the reduction degree is high, is mainly caused by the reduction processes during operation. The situation in the peripheral zone is that the reduction degree is low and in this situation the breakdown is mainly caused by the long residence time of materials around 500 deg C during blowing out of the furnace. This was illustrated with the dissection results for the centre working Hirohata BF no.1. Another factor in maintaining the size of the burden materials is that in the central region of Hirohata BF no. 1, cracks if generated fused immediately because of the high temperatures and the rapid reduction taking place. Degradation is generally a problem having maximum concerns with sinters. Examination of the size distribution of pellets revealed that they were hardly pulverized and maintained their original shape.
Influence of gas flow – To further prove that the determination of the shape of the cohesive zone is by the gas flow within the furnace, core samples were taken from the Hirohata BF no.1 and Kukioka BF no.4 and their permeability was determined. Then their permeability was related to gas flow and gas velocity distribution profiles were prepared. These profiles can be directly related to the softening-melting layer distribution. The gas flow in the lower part of the BF is fast, 7 m/sec to 9 m/sec but slows considerably in the softening-melting layers to 2 m/sec to 4 m/sec thus indicating the poor permeability of the softening-melting layers. As the gas ascends the shaft its velocity naturally decreases due to the drop in gas temperature.
Cohesive zone control
It has been shown that the shape of the cohesive zone varies from BF to BF and much attention needs to be given for its control. The control of the cohesive zone is very dependent upon burden distribution. For maximum production, at the expense of fuel rate, a strong centre working profile is to be adopted, but if the fuel rate is to be minimized, then a less centre working practice is to be followed. Indeed, this is very much visible when comparing a strong centre working furnace, like Hirohata BF no.1 with moderate centre working furnace, like Kukioka BF no. 4. This point can be well explained by relating productivity to the height of the cohesive zone above the tuyeres (Fig 4). The higher the position of cohesive zone in the furnace, the greater is the productivity, although at the expense of an increase in fuel rate.
Another point concerning control of the cohesive zone is its effect on the refractory lining. If the wall temperature of the furnace is too high, then refractory wear is appreciable and one can expect a reduced life of the BF. Thus, for maintaining the refractory thickness, it is necessary to control the cohesive zone so that the wall temperatures are maintained at minimum levels.
Fig 4 Schematic profiles of softening-melting zones and relationship BF productivity and height of the melt-down zone
The role of S in the melting process is governed by the Fe-S-O phase diagram. There is a necessity of a reaction between solid metallic iron and wustite in the burden with gaseous S, in the ascending gases. These phases react to form a eutectic of chemical composition 24 % S, 9 % O2, and 67 % Fe, having a melting point of 915 deg C. Once formed this liquid gains temperature as it descends the furnace, dissolving solid metallic iron and wustite which cause a change in liquid composition along a path until at certain point, the liquid splits into two conjugate liquid phases. Further increases in the temperature cause first part of the liquid to dissolve more solid iron, moving its composition along a path while the second part of the liquid dissolves more iron oxide and moves along the another composition path. Thus there are two phases (i) a liquid metal phase, and (ii) a liquid slag phase. The presence of silica in the system does not appreciably alter this mechanism. Indeed it moves the miscibility gap. Hence the separation of the nascent liquid into liquid metal and liquid slag phases occur at lower temperatures.
Once formed the two liquids go their own separate ways. The liquid metal dissolving solid iron, C and S become the final metal phase. The slag during its descent dissolves alumina, silica and lime from the coke ash, burden gangue and fluxes to form the final slag phase. A study has also shown that that the presence of hydrogen sulphide, in a CO / N2 gas mixture, lowered the melting point of iron ore sinters and pellets due to the formation of the liquid Fe-S-0 phase.
Alkalis are also thought to be closely associated with the initial melting process in the BF. Study with regards to the distribution of alkali, shows that the alkali is concentrated in the softening-melting layers. The reason for this is that alkali compounds, inherent within the burden and coke charged into the furnace are reduced and at temperatures in excess of 800 deg C to 900 deg C, the alkalis vapourize, as a metallic element or as a cyanide, and are swept into the softening-melting layers where they concentrate . As the softening-melting layers descend the alkali evaporates and continues the cycle.