Smelting in cupola furnace for recarburization of direct reduction iron (DRI)

José L. Enríquez, Enrique Tremps, Iñigo Ruiz-Bustinza, Carlos Morón, Alfonso GarcíaGarcía, José I. Robla, Carmen González-Gasca E.T.S. de Ingenieros de Minas, Dpto. de Ingeniería de Materiales, Universidad Politécnica de Madrid (UPM). C/Rios Rosas, 21 28003 Madrid, España E.T.S de Edificación, Dpto. de Tecnología de la Edificación. Universidad Politécnica de Madrid (UPM). Avda. Juan de Herrera, 6 28040 Madrid, España Centro Nacional de Investigaciones Metalúrgicas (CENIM). CSIC, Avda. Gregorio del Amo 8, 28040 Madrid, España Universidad Europea, Laureate International Universities, C/Tajo s/n Villaviciosa de Odón, 28670 Madrid, España *Corresponding author: joseluis.enriquez@upm.es ABSTRACT


INTRODUCTIONTOP
The cupola (American Foundrymen's Society, 1965) is a shaft furnace whose primary characteristic is its height. (Fig. 1). Thermically it is a counter current heat exchanger. During combustion, air enters through tuyeres and burns the coke inside, releasing heat (Enríquez and García, 1993). The iron introduced in the charging door is heated and melted by rising combustion gases. The melted metal drips through the coke bed and falls to the hearth, draining out through the tap hole. The high level of thermal efficiency of this type of furnace allows the melting of any material. Chemically, cupola is a continuous reactor. The iron mixes with CO from gases given off by the burning coke, melts, and carburizes. The melted iron drips through the burning coke bed, absorbing more carbon along the way. The third role played by coke is mechanical. It supports the weight of the melting bed because it is the only component introduced into the furnace which remains solid at the inner temperatures of operation.

Carbon contentTOP
It is a function of the composition and temperature of the slag, coke-liquid interfaces, and coke characteristics (Tartera et al., 1987;Alvarez et al., 2002). In practice the maximum theoretical value is not reached. By recarburizing iron, quality cast irons are made from unsaturated inexpensive materials.
The solubility of carbon in iron is given by Neumann's formula (1) (Neumann, 1968): where, N C is the mole fraction of carbon.
In addition: The influence of diverse elements on the solubility of carbon in the liquid iron is expressed in the following way: Experience shows that some elements (V, Cr, Mn, Mo) favour the solubility of carbon in iron, while others, especially Si, P and Al, strongly diminish it (Neumann, 1968). Figure 2 shows the effect that silicon content and iron temperature have on carbon solubility. We can see that the solubility of carbon in iron grows with the iron temperature, and decreases with an increase in silicon content. Another way to express carbon solubility in iron is with the formula (4) (Shibaev and Grigorovich, 2008): More recently Janerka (2010) calculated recarburization in the following way: where E is the recarburization efficiency (%); M M the metal mass (kg); C F the carbon content at the end of the process (%); C I the carbon content at the beginning of the process (%); M R the mass of the recarburizer (kg) and C R the carbon content in the recarburizer (%).  The "carbon equivalent," or CE, is the sum of the metallurgical effects of total carbon and of the most frequent elements (Karsay, 1980;Creese and Healy, 1985;Heine, 1986): Empirically (Enríquez and Hernáez, 1982;Coon, 1983), we know that the absorption of carbon in the cupola furnace depends on the composition of the base iron; the characteristics and operation of the furnace and the coke quality.
Sulphur and coke ashes work against recarburization. We know that operating temperature, height of the hearth, coke size, and coke bed height favour it. The fluxes, metallic oxides and lining oxides make up slag which combines with coke ashes improving the reactivity of the coke. Experience shows that factors such as blast enrichment with oxygen, hot blast, balanced blast, or the addition of fluxes increase the value of carbon capture by iron; Si and P decrease it (Levi, 1947;Leyshon and Shelby, 1972;Katz and Landefeld, 1985).

Silicon contentTOP
Silicon in the iron decreases by oxidation and increases as a result of the reduction of silica from the refractory, coke ashes, slag or gangue of DRI by the carbon present in the coke and iron. The reactions are: Si+2FeO↔SiO 2 +2Fe (7) SiO 2 +2C↔Si+2 CO Slag, which is basic, tends to combine with acid oxides of some elements (silica SiO 2 , alumina Al 2 O 3 ) taking them out of the equation and changing the equilibrium. As a consequence, the first reversible reaction (oxidation of silicon) will be forced towards the right, while the second goes towards the left. In this way silicon, which as we have seen works against recarburization, is removed, with a higher degree of carbon content in the molten iron. Figure 4shows the effect of the percentage of coke between charges (weight of coke in every charge divided for the respective weight of metal) and the blast temperature, on silicon capture. Considering the above, it can be concluded that the absence of silicon in the cold charge materials is key to recarburization, due to its negative effect and its abundance (if it is present) in the melt.

PREREDUCED PELLETSTOP
The use of prereduced pellets (Direct Reduction Iron, DRI) has increased because they: a) resolve situations where blast furnace coke for pig iron is scarce and b) dilute contaminating elements which cannot be eliminated in conventional operation.
The classic charge of a foundry furnace is composed of pig iron, cast iron scrap, steel scrap, and ferroalloys. Recycling scrap causes contaminants to accumulate in produced cast irons. It would be better to use pig iron, but that would be significantly more expensive. DRI pellets, almost pure iron, transform in cast irons without any undesirable elements (Fernández and Mendiola, 1977;Lemus et al., 1998). The present project looks at using this material as raw material in foundry shops (Dahlmann and Hussmann, 1976;Pietsch, 1976;Enríquez and Tremps, 1995). There are three stages in this project: chemical and physical characterization of the pellets; experimental melting in a pilot furnace and experimental melting in an industrial furnace.
The Carter Basicity Index (I B ) is the ratio of basic oxides and acidic oxides in a gangue or slag such that: It can be said that: -A low index (high SiO 2 ) means acid slag, reduction of silica and the passing of silicon to the melt and, consequently, poor recarburisation.
-This acid slag demands additional feeding of basic flux to neutralize it and, logically, more energy to melt that flux.
-For this reason the correct operation should be hot and basic, with tall coke beds and a high percentage of coke between charges, a very hot blast, and, if is possible, a basic or neutral refractory lining (magnesite, dolomite or graphite).

Physical analysisTOP
In the sizing, 85% of the material has less than 15 mm mean diameter and shows a good homogeneity of size. However, this seems small if they are falling among pieces of coke larger than 100 mm diameter. Physical measures are shown in Table 1, with resistance measurements also. These values lead to the conclusion that the material is quite durable, resistant to crushing and the weight of the successive charges it must support.

MELTING IN EXPERIMENTAL CUPOLA FURNACETOP
Melts have been made in two cupola furnaces, first on a small pilot experiment, and then on an industrial scale furnace (Enríquez, 1975;Geck and Maschlanka, 1976;Enríquez and Crespo, 1998

RunsTOP
Nine runs were done, each of 1000 kg (40 metallic charges), with variable proportions of DRI and ingot per charge. The process followed for melting the iron was the same as in previous industrial runs with conventional charges, starting with a preheating blast (10 Nm 3 min −1 ) and continuing with the total blast (15 Nm 3 min −1 ) before initial tapping. Each metal charge weighs 25 kg; the charge of coke, 5 kg and the charge of flux limestone, 1 kg. Optical pyrometer is employed for the measure of liquid iron temperature in tap-hole. The productivity is about 800 kg h −1 . The tilting forehearth was preheated to 1000 °C by means of a propane burner.

Furnace behaviourTOP
In contrast to conventional heats in the same furnace (Chastain, 2000), the following empirical observations can be made: -Yellow oxidizing flames passing in front of the charging door.
-Smaller blast rate and greater back pressure.
-Great amount of a black oxidized slag (the normal one must be gray and vitreous).
-Low temperature of tapped iron.
-The wear on the refractory lining does not appear to vary. Table 2 gives the values of the slag for different percentages of pellets in the charges. The column at the right shows the basicity index (I B ).

Composition of obtained meltsTOP
Determined by emission spectrometry (Enríquez, 1981a;Enríquez and Echarri, 1988). The values of carbon and silicon for each melt sample can be seen in Table 3, as well as those for the Carbon Equivalent (CE). In the column to the right, average values are given. Carbon self-regulates in the furnace, unlike silicon which depends on charges of ferrosilicon.

Discussion on the pilot furnaceTOP
Increasing the percentage of pellets decreases the basicity index (I B ) of the slag from 0.31-0.46 for 25% pellets in the charges to 0.10-0.18 for 100%. As a result its desulphurating power diminishes. Higher acidity (silica content) in the slag, also, favours the reduction into free silicon and slows down, consequently, the carbon picks up.
It has a higher FeO content and oxidizing atmosphere that affects the fluidity and recarburization of the molten iron negatively. The iron oxide content increases from 1.4-9.5% for 25% pellets into 12.0-36.2% with 100% pellets.
In regards to the composition of obtained cast iron, the degree of recarburization was not satisfactory.
The average values of carbon content are in the range of 1.85-2.90%. The goal was to reach values of 3.00-3.50% of carbon, range of composition for industrial cast irons.
Summarizing, the basic desired objective has been obtained, which is, the inexpensive synthetic production of saturated iron-carbon alloys in a cupola furnace. However, the results have been mediocre. It cannot be known with certainty whether the difficulties that were found were due to the small size, the consequent difficulty of operation of the cupola furnace, or the use of the product (DRI) which was slightly metallized with a lot of gangue and was small as basic component in the charge.

INDUSTRIAL FURNACE MELTINGSTOP
The difficulties found using the experimental pilot furnace led to further work with an industrial sized furnace which was retired because it was substituted with better, more modern units (Hornung, et al., 1975;Enríquez, 1981b;Enríquez, 1982;Enríquez and Alonso, 1996).
The furnace is in a foundry shop for automotive parts. It has an inner diameter of 600 mm, acidic refractory, cold blast, charge by means of tilting skip, and discontinuous tapping of iron and slag. The productivity is about 2200 kg h −1 . A monorail connects it with a 5 t channel induction furnace (Enríquez, 1980;Enríquez, 2000) that holds, homogenizes and overheats the liquid iron produced in the cupola.
Although it was not the optimal solution for the process being studied, more positive results than those obtained with the small pilot furnace used in the former stage of the project are hoped for.

MeltingTOP
Two melting runs were done, using the same routine. In the first 5700 kg (19 charges Table 4.

Operation characteristicsTOP
-The operation of cupola was normal, with no anomalies worth mentioning. The first metal was able to be cast without any temperature problems.
-During the run, the temperature of liquid iron, measured with optical pyrometer, was 1390 °C.
In the pilot cupola it was only 1370 °C. The molten iron showed a satisfactory fluidity.
-The volume of slag was greater than that usually produced with conventional charges but it did not interfere with the operation.
-Observation through the peep-holes of tuyeres showed the free fall of intact cold pellets between the coke pieces. This circumstance could be disadvantageous for the carburisation and temperature of the molten iron.

Observation of obtained slagTOP
The slag has a good quality, with satisfactory fluidity. It shows a green colour and vitreous appearance that is characteristic of the slag from the cupola with acid lining. The composition is given in Table 5.
The basicity index falls from 0.27 to 0.09 when the amount of DRI increases from 0 to 50%. With the same proportion, the tenor of FeO increases from 1.7 to 15.2%. These values are approximately the same as those obtained in the pilot furnace.

Composition of obtained meltsTOP
Analysis values of samples taken along the run are given in Tables 6 and 7 for 33% and 50%, respectively, of DRI in the charges. The average values for carbon were 3.33 and 3.34. They can be considered satisfactory. The results for equivalent carbon CE were 3.940 and 3.907, which was clearly hypoeutectic.  those in the small pilot cupola but slightly poorer those obtained during conventional runs with carburised materials in industrial cupolas. Going from 33 to 50% pellets per charge did not result in a substantial variation in average carbon or CE values and the furnace self-regulated spontaneously.

DISCUSSIONTOP
-Silicon. The furnace was part of an industrial factory and the materials obtained had to be suitable for the automotive parts produced there (iron -carbon -silicon alloys). Occasional variations in the content of silicon found from one tap to another may be the result of the little pellets falling across the hollow spaces between coke pieces. It must be borne in mind that taking samples from tap hole flow gives erratic composition values which are then homogenized in the holding furnace.
-Sulphur and phosphorous. They are not seen; notable variations, such as the sulphur present come only from the coke. Phosphorus is in the gangue and does not vary substantially either. A more basic slag should be favourable for attaining a better degree of desulphurisation.
In any case, the amounts of both are well within permitted limits for quality casting for automotive castings owing to the scarce content of these elements in the DRI.
-Alloy elements. Variations in different samples can be seen which, as in the case of silicon, are due to the premature descent of the pellets. Analysis of successive samples from the holding furnace indicate the gradual disappearance of copper and molybdenum from the mixture contained in this furnace, after being diluted by melted iron falling from the cupola tap hole.
This is important to consider when it is necessary to decrease the amount of trace elements which cannot be entirely eliminated in conventional operations.

CONCLUSIONS AND RECOMMENDATIONSTOP
-Cupola furnace, in order for this unit to operate satisfactorily, it has to have a usable diameter at least 700 mm which has a productivity of P=6 D 2 ≈3000 kg h −1 . Smaller sizes cause manipulation difficulties in molten metal and loss of temperature.
-The dimensions of the small experimental pilot furnace are clearly below, which does not make the conclusions that are deduced from its use very reliable. The dimensions of the cupola furnace used in this work approach the minimal prescribed value.
-As for the metallic charge materials, the basic rules of operation of cupolas require that the size of the pieces must be equal to 1/3 of the internal diameter of the furnace in the melting zone.
That is, 130 mm in pilot furnace and 200 mm in industrial furnace. The DRI pellet used (there were no others) have a diameter of 15 mm, well below the established standards, despite which it has been working with them and get acceptable products.
-There is a considerable increase in the quantity of acid slag, due in large part to the gangue of the pellets. It would be better to use pellets with less gangue and not exceed the 30% DRI in the furnace cold charge.
-The filtration and fall of pellets detected gave us snapshot glimpses with significant variation in the composition and temperature of iron in the tap hole. This heterogeneity is greatest when the furnace cold charge consists of 50% of pellets DRI. This could be avoided if the pre-reduced iron was canned or briquetted.
-Various furnace characteristics employed (cold blast, tilting skip basket, intermittent casting) are not the best solutions with pellets.
-The use of a channel induction furnace as a holding furnace, resolve partially the problems of composition and temperature heterogeneity mentioned earlier.
-Acquired experience shows that it is inadvisable, under current operating conditions, to use more than 50% pre-reduced iron pellets in the cold charge. The maximum suggested would be 33%.
-More efficient operation could be achieved by using hot blast and basic lining in the cupola.
Continuous and self slaged tap hole require an induction fore hearth to regulate melt temperature and composition (duplex cupola -induction).