Desarrollo para la producción de acero con bajo contenido de fósforo en las operaciones de Arcelor Mittal Tubarão

The growing demand to produce steel with lower phosphorus (P) content happens in tandem with the increase in the content of this element in the ores used in the production of pig iron, leading to a constant evolution in the process of steel dephosphorization in BOF converters. Arcelor Mittal Tubarão (AMT), located in the municipality of Serra, Brazil, currently produces 5 million t of steel, but has an installed capacity of 7.5 million. This work aims at showing the development carried out in the dephosphorization of silicon-aluminium killed steel produced in the Arcelor Mittal Tubarão converters. The analysis of process variables such as flux addition, oxygen lance position and temperature at the end of blow are based on classical phosphorus partition models. The results compare phosphorus values in liquid steel before and after modifications in the variables and the refractory wear caused by the new procedures applied to AMT steelmaking converters .


INTRODUCTION
In most steel grades, the phosphorus is considered an undesirable element, which must be reduced as much as possible.The most common way to reduce this element in fully integrated steel-making companies is during the oxygen blowing process in the converters.This paper presents the production development of silicon-aluminium killed steel with phosphorus content below current standards in the final product (slab).Therefore, the purpose of this study is to evaluate the steel dephosphorization process performance of Arcelor Mittal Tubarão converters to meet the demand for low phosphorus, lower than 0.015 % for silicon-aluminium killed steel with manganese higher than 1.0 % in the final product.
Several authors referred to in this paper have studied the process of steel dephosphorization, working with the phosphorus partition coefficient (Lp) at the end of the blow, where (Lp) = the correlation between phosphorus content in the slag divided by the percentage of this element in steel.
The dephosphorization reaction could be considered as [1 and 2] : Balajiva [1] suggests a correlation between the phosphorus partition coefficient (Lp) with CaO (lime), the percentage of total Fe in slag and temperature, i.e.: log lp = 11.80 log (% CaO) -5 log (% Fe t ) -C (2) where C, is a constant for a given melt temperature: C = 21.13;21.51 and 21.92 for the melt temperatures of 1550 °C, 1585 °C and 1635 °C respectively.The elements or compounds in parentheses are present in the slag and the elements in brackets are dissolved in the steel.

22350
-----16 ± 0.4 T (% P) where: lp = ---(% P) Considering equations ( 3) and ( 5), it could be noticed that an increase in basicity, the iron content in the slag and the reduction of temperature enhance the dephosphorization process, as observed by the increase of the phosphorus partition coefficient.
Turkdogan [4] proposed a correlation of the phosphorus partition coefficient, considering the reaction (6) with a new variable which aims at measuring the slag basic oxides (BO), as follows: (% P) where: lp = ----[% P] 21740 log(lp) = 0.071BO + 2.5log [% P] + -----9.87(7) where BO is defined in equation (8): Equation (7) intends to provide the capacity to absorb phosphorus from the slag in terms of basic oxides, where CaF 2 has the same weight as CaO and MgO is 1/3 of the CaO weight.Therefore, fluorspar, as well as the reduction of slag viscosity increasing the reaction kinetics have the same effect in improving the thermodynamic conditions as the CaO [4] .
Notice that an increase in basic oxides and a fall in temperature both decrease the activity coefficient of P 2 O 5 favoring dephosphorization.Turkdogan and Pearson [5] proposed that the Raoultian activity coefficient "g FeO " decreases with increasing binary slag basicity (B = % CaO / % SiO 2 ), as shown in table I.
According to the author, a decrease in the "g FeO " makes the Raoultian activity of FeO "a FeO " also decrease, which causes the reaction (4) to occur towards the decomposition of CaO P 2 O 5 , which would result in a reduction of the slag capacity to retain phosphorus.
Turkdogan [4] affirms, however, that the slag basicity is not the most appropriate parameter to describe the dependence of the phosphorus absorption capacity of the slag."BO" would be more important.
However, according to Balajiva [1] , there is an optimum relation between the % FeO in slag, basicity and the phosphorus partition coefficient (Lp) between slag and metal, as shown in figure 1.

EXPERIMENTAL
In order to develop low phosphorus steel, 12 heats were produced.Changes in the process were based on some of the parameters mentioned before, with focus on variables which enhance the removal of phosphorus during the production process.The methodology was based on current results of phosphorus content in normal grades of steel, changing the process variables in order to increase the phosphorus partition coefficient.
For the production of steel in development, the phosphorus content must be around 50 to 60 ppm lower than the phosphorus specification of the final product, which is of 150 ppm, due to the contamination that may occur from the slag carried over from the converter heat tapping to the ladle, and also from the pick-up of phosphorus content in the Mn-Fe-alloy, which is usually added to these steel grades.
The main actions in the process variables in order to achieve the phosphorus partition are described below.
[% P] coefficient, oxidized iron is the variable that affects positively the removal of phosphorus.Thus, it was aimed to achieve lower carbon content in the end of the blow.In fact, oxygen mass balance calculations were made in order to obtain an excess of blown oxygen volume, which allows higher oxidation of carbon, silicon and manganese, also a larger amount of iron oxide in the slag, as it can be seen in table II.
The effect of increasing the amount of flux used during the oxygen blowing was also analysed based on the Turkdogan [4] model (Ec.( 7) and ( 8)).In this case, the CaO quantities were increased by adding lime, and also fluorspar (CaF 2 ) to speed highly basic slag formation.Regarding Si-Al killed steel grades with high manganese, the increase in lime was, on average, of 3 kg/t of steel produced and the fluorspar was of approximately 1 kg/t of liquid steel.The average net result (Table II) was an increase in basicity (% CaO / % SiO 2 ).
Temperature is a highlighted item in all the models showed before and it is inversely proportional to the phosphorus partition coefficient, i.e., lower temperatures enhance the phosphorus removal from the steel.Despite this, high additions of manganese alloys in the bath after steel tapping push the temperature to higher values compared with current grades of steel, to compensate the heat loss.Thus, it was not possible to reduce the temperature values at the end of the blow.
Table III shows the summary of the actions taken in process variables and the impact on the partition coefficient based on the mentioned authors [1][2][3][4][5].All the changes shown in table III were performed simultaneously on all 22 heats.Throughout the production process, all fluxes and materials are added by weighing on certified scales and the data are recorded in the computers in the process.All the process information presented in this report was collected from this database.
Temperature measurements and steel samples were collected by using probes connected to a sub lance which is inserted in the liquid steel bath during and at the end of the blowing.The phosphorus content in steel was analysed at the AMT chemical laboratory by optical spectroscopy and slag samples were analysed using an X-ray diffractometer.

RESULTS AND DISCUSSION
The results of the phosphorus evolution from hot metal to the final product (slab) in the different experimental heats, after the changes, shown in table III, are shown in figure 2. As it can be seen, there was a strong removal of phosphorus in the stage of oxygen blowing and after a phosphorus reversion in the following process.That occurred because of the slag carried over during the tapping process and from the pick-up of residual phosphorus of Fe-alloys added.The maximum phosphorus reversion for high Si-Mn-Al killed steel was observed to be of 60 ppm.The results of the average values in the end of the blow and in the slab of current and experimental heats, are shown in table IV.
The phosphorus removal efficiency in the experimental heats is compared to the current ones through figure 3. Actually, it is observed that in general, phosphorus partition coefficients (Lp) were higher than the current ones, endorsing the plan of action reported in table 2. This result is due to an increase of % FeO in the slag, the higher addition of CaF 2 , as well as to an increase in the oxygen available to form P 2 O 5 , as shown in table II.All these actions  Figure 4 shows the distribution of phosphorus content at the end of blow of current and experimental heats.As can be seen in the figure, phosphorus occurrence frequency is lower than 0.015 % was always higher than the experimental heats, and the occurrence of phosphorus above 0.015 %, was always higher than the current ones.These results show that the process changes, according to table III, were effective to ensure results below or equal to 0.015 % of phosphorus content in the final product (slab).This occurred due to the same reasons aforementioned.Despite the good results of the tests regarding the phosphorus content at the end of the blow for experimental heats, the converter refractory wear index, evaluated from bricks thickness measurements, was significantly higher than in the current heats, as it can be seen in figure 5.This undesirable effect is due to the higher FeO content in slag, and also the increase in fluorspar addition.Both procedures aim to decrease slag viscosity, which increases refractory wear.High temperatures at the end of the blow also contribute to increase refractory wear.

CONCLUSIONS
-The dephosphorization results of experimental heats were significantly higher than the current ones.-Higher phosphorus removal in the experimental heats is mainly due to: An increase of oxidation level of the system; Changes made in the process, such as the increase of oxygen volume and blow pattern modification (changes in the oxygen lance height), were effective for this item.Higher additions of fluxes; The increase in lime and fluorspar additions were also satisfactory in the achieving the expected results.-These results are consistent with the available models proposed in the literature describing the phosphorus partition coefficient equilibrium, as shown in this paper.-The side effect of increasing the efficiency of phosphorus removal was the raise of the refractory wear rate in the converters, due to an increase of slag oxidation, the presence of fluorspar, both associated with higher temperatures at the end of the blow.

Figure 2 .
Figure 2. Phosphorus content evolution from hot metal to slabs in the experimental heats.
associated decreased slag viscosity and enhanced dephosphorization kinetics.On the other hand, an increase in the addition of CaO, supports the dephosphorization thermodynamically, by reducing the P 2 O 5 activity, also offering a higher amount of CaO to fix the P 2 O 5 in slag, according to reaction: P 2 O 5 + 3CaO = P 2 O 5 .3CaO.The results reassure the relation of empirical models to describe the phosphorus partition coefficient, despite the wide dispersion when compared to industrial data.

Table I .
Effect of slag basicity on FeO activity coefficient

Table II .
Average compositions values of hot metal and final slag Tabla II.Valores medios de la composición del arrabio y escoria final

Table III .
Summary of actions in the process and its impact on the phosphorus partition coefficient Tabla III.Resumen de las acciones en el proceso y su impacto en el coeficiente de partición de fósforo

Table IV .
Average values in the end of the blow and in the slab of current and experimental heats * Predicted by mass balance; σ: standard deviation.