Evaluation of the Furnace Method for the production of low carbon ferrochrome

Abstract

Low carbon ferrochrome is a primary alloying element in the production of stainless steel. The Mixing Method (Perrin and Duplex) processes are most commonly used for the production thereof. However, during the cocktailing step, process temperatures are extremely high, resulting in rapid deterioration of the ladle refractories and fuming of the products. The high temperature is a result of the exothermic silicothermic reduction reactions. A significant amount of energy is lost to the atmosphere. The Furnace Method, and in particular the Liquid Feed Furnace Method, has the potential to improve on the shortcomings of the Mixing Method. A techno-economic evaluation was performed on the different process routes to identify whether or not there is merit in choosing one over the other. The evaluation showed that the capital cost for the methods are comparable, but a saving in the operating cost is achievable when using the Furnace Method instead of the Mixing Method. Savings of 7.3 to 7.9% were calculated for the Solid Feed Furnace Method and 9.6 to 10.7% for the Liquid Feed Furnace Method, when compared to the Mixing Method. This is largely due to the lower energy requirement and raw material consumptions for the Furnace Method. The oxidising conditions in the Mixing Method ore-lime melt furnace, combined with a high slag basicity and high operating temperatures, are very conducive for producing hexavalent chromium, which is severely toxic to humans. Leaching tests, performed on a dust sample from an existing facility that uses the Perrin process, confirmed that a significant amount of Cr(VI) was produced. This poses a severe health and safety risk, as well as a financial burden to properly neutralise and dispose of any Cr(VI) that has formed. A closed furnace with little to no air ingress is used in the Furnace Method, thereby ensuring a more neutral atmosphere. A lower basicity slag, with a lower liquidus temperature is also used, thereby further decreasing the amount of Cr(VI) that would be likely to form. Having established that the Furnace Method has definite advantages over the Mixing Method, the preferred refractory-slag system was identified by means of thermochemical modelling. Different combinations of lime and doloma fluxed slag were modelled with magnesia and doloma refractories. Quartz fluxed slag in contact with an alumina lining was also considered to identify any unexpected benefits. The required slag liquidus temperature, to be compatible with low carbon ferrochrome alloy (which has a very high liquidus temperature), was determined to be between 1550 and 1700ºC. Modelling showed that a quartz fluxed slag had a liquidus temperature well below this range for a very wide basicity range evaluated. Such a slag would therefore not be suitable. While a doloma fluxed slag had a liquidus temperature within this range, the slag basicity for that range would be very low and not be compatible with any of the two basic refractory systems evaluated. Doloma refractories were also found to suffer severe wear at an operating temperature of 1750ºC, the temperature required to ensure that the alloy is molten. The required basicity range for a lime fluxed slag that had a liquidus temperature between 1550 and 1700ºC was found to be 1.68 to 1.90. The slag only became saturated in MgO above a slag basicity of 1.99. However, in a commercial-size furnace, the temperature at the refractory hot face would be lower than the process temperature, provided that the slag bath was not excessively turbulent. Sidewall cooling would also assist in maintaining a slag freeze lining. Operating with a slag basicity within the 1.68 to 1.90 range should therefore not pose a severe risk. As the process will be operated on a semi-batch basis, the variation of the slag and alloy composition throughout the heat could potentially result in a material that is not compatible with the refractory lining. Two scenarios were modelled, where FeSiCr was added to an ore-lime mixture, and vice versa. The first was found not to be detrimental to the refractories, but there is a concern regarding the high liquidus temperature of the ore-lime melt at the start of the heat. For the second scenario, magnesium was found to report to the gas and alloy phases at the start of the heat. This is of great concern as it would severely damage the refractory lining. To utilise as much of the exothermic energy as possible, while ensuring the integrity of the refractory lining, it was suggested to feed a portion of the ore-lime mixture first, followed by all of the FeSiCr before feeding the remainder of the ore-lime mixture. Feeding crushed, solid FeSiCr allows for better control. However, the latent heat and sensible heat of the FeSiCr would not be utilised. Smelting tests were performed with samples that had slag target basicities of 1.5 and 2.0 to investigate the phases that were formed and the severity of the slag-refractory interaction. The composition of the slag with a basicity of 1.5 corresponded well with that predicted by the thermochemical model, while the slag composition for a target basicity of 2.0 was very variable. This was due to the high solids content in the slag, which was operated close to its liquidus temperature. The slag CrO content was much lower and the alloy Cr content much higher for both tests, when compared to the values predicted by modelling. The extent to which the reduction reactions occurred was therefore higher than predicted by modelling. The reasons for this would have to be verified by analysing the mechanism by which the reactions occur. The complete dissolution of a refractory disc that was placed in the mixture for the test with a slag target basicity of 1.5, along with slag penetration into the high density magnesia crucible, indicates that the 1.5 basicity slag was not compatible with the magnesia. This is in agreement with the model. Although the refractory disc was still discernable for the test at a slag target basicity of 2.0, it was severely worn. Slag penetration into the high density magnesia crucible was also evident. A freeze lining would therefore always be required. The Furnace Method can be considered as a suitable alternative to the Mixing Method. However, care should be taken to control the slag basicity in the region of 1.70 and a freeze lining should be maintained to protect the refractories. The batch feed sequence is also critical to find a balance between having a liquid slag, while still retaining refractory integrity.