Modelling of freeze layer formation and refractory wear in direct smelting process
The work discussed in this thesis is aimed at examining the formation of freeze layers and refractory wear on water-cooling elements within direct smelting processes through the use of computational modelling techniques. The motivation of performing this work is to examine the cooling of regions of the Smelt Reduction Vessel of the HIsmelt process closer to the molten bath material. HIsmelt is a novel process for the production of pig iron which has been under development by Rio Tinto and is now being ommercialised. The previous work performed in this are has been reviewed with particular focus on the refractory wear mechanisms as the solidification algorithms have been thoroughly implemented within the Computational Fluid Dynamics (CFD) framework PHYSCIA used within this work. The governing equations along with the Finite Volume discretisations of these equations are set out within this thesis. Some comment is made about the solution methods used, and how boundary conditions are implemented. The Free-surface flow and Solidification governing relationships are also described as these are important for investigating the formation of freeze layers. The implementation of the refractory wear mechanisms are discussed in some detail. The three mechanisms implemented are for the penetration of slag into the refractory, the corrosion of the refractory by this penetrated slag; and the erosion of the refractory by the bulk flow of slag within the furnace. To be able to reasonably predict refractory wear, it is necessary to make the properties of the materials within the system temperature dependent. During the pilot plant trials at the HIsmelt® Research and Development facility, located in Kwinana Western Australia, accretions formed on the end of the solids injection lances. These accretions have been termed Elephant's Trunks. With the imminent construction of the Development Plant which injects the iron bearing feeds at an elevated temperature rather than at ambient temperatures used on the pilot plant, the formation of these pipe-like accretions under both the cold and hot injection conditions have been examined. This work provides confidence that the freeze layers predicted from the model will reflect those formed within the furnace. To evaluate the effectiveness of the refractory wear mechanisms, data from experimental and the HIsmelt pilot plant have been modelled. Sections of refractory samples from an induction furnace test and a rotary slag test have been modelled. The results are in agreement with the profile and affected regions of the sectioned refractory test pieces. A part of the HIsmelt pilot plant Smelt Reduction Vessel (SRV) has been modelled for the period of campaign 8-1 & 8-2 (just over 20 days). The predicted wear is in agreement with the measurements taken after the vessel had been cooled. To bring together freeze layer formation with the refractory wear mechanisms, a water-cooled element was modelled for the sloping slag section. The results show the growth of a small freeze layer that is consistent with the small freeze layer seen on the upper cooling panels of the pilot plant SRV. This model is an ideal tool to evaluate different water-cooling strategies for HIsmelt and other similar direct smelting processes. This work has developed models that predict the formation of freeze layers and refractory wear within direct smelting processes. The models have focused on slag-refractory interactions and further work would be needed to extend the refractory wear models to account for metal-refractory interactions. To examine spalling, stress calculations could be performed to determine when this may occur.