Supersonic oxygen jet is used in both Basic Oxygen Furnace (BOF) and Electric Arc Furnace (EAF) steelmaking to remove impurities from the liquid iron by oxidation reactions. In EAF steelmaking, the oxygen jet is injected between 35 to 45 degrees from the vertical position whereas in BOF steelmaking the oxygen jet is injected through a vertical lance from the top of the furnace. The understanding of the supersonic jet behaviour, jet-liquid interaction and droplet generation rate from the liquid surface inside the furnace is important in optimizing the steelmaking process. In the present study, the Computational Fluid Dynamics (CFD) technique and physical modelling was used to study the supersonic jet behaviour, jet-liquid interactions and splashing phenomenon inside the steelmaking furnace. A CFD model was developed first to investigate the effect of high ambient temperature on the supersonic jet because in steelmaking the ambient temperature is between 1700 to 1800 K. The results, when compared with available experimental data, showed that the k - ε turbulence model, with compressibility terms, underpredicts the potential core length of the supersonic jet at high ambient temperature. A temperature corrected k - ε turbulence model was proposed in this study that decreases the value of Cμ at higher temperature gradient which in turn reduces the turbulent viscosity. As a result, the growth rate of turbulence mixing region decreases at higher ambient temperatures which increases the potential core length of the jet further. The results obtained by using the modified turbulence model were found to be in good agreement with the experimental data. The CFD simulation showed that the potential core length at steelmaking temperature is approximately 2.5 times longer compared to that at room ambient temperature. It was then followed by the CFD modelling of coherent supersonic jet where the supersonic jet is shrouded by a combustion flame which results in longer potential core length for the coherent supersonic jet compared with normal supersonic jet. The combustion flame was created using CH4 and O2 as fuel and oxidant respectively. The CFD model showed that the shrouding combustion flame reduces the turbulent shear stress magnitude in the shear layer. As a result, the potential core length of the jet increases. The potential core length of the coherent supersonic jet was also found to depend on the gas used as the central supersonic jet. The numerical results showed that the potential core length of the coherent supersonic oxygen and nitrogen jet is more than four times and three times longer, respectively, than that without flame shrouding, which were in good agreement with the experimental data. After modelling the gas phase, another CFD model was developed to simulate the jet liquid interaction. A new approach was proposed where two computational domains were used to avoid the difficulties that arise from the simultaneous solution of compressible supersonic gas phase and incompressible liquid phase. The effect of shrouding gas flow rates on the axial jet velocity distribution, depth of penetration and velocity distribution of liquid phase were investigated. In this case, only compressed air was used to shroud the main supersonic jet instead of a combustion flame. A higher shrouding gas flow rate was found to increase the potential core length, depth of penetration and liquid free surface velocity which in turn contributes in reducing the mixing time. The CFD model successfully predicted the formation of surface waves inside the cavity and consequent liquid fingers from the edge of the cavity which were experimentally observed by the previous researchers. An experimental study was carried out to investigate the effect of different operating conditions (lance angles, lance heights and flow rates) on the wall splashing rate. Air was injected on water surface in a 1/3-scale thin slice model of the steelmaking furnace where dynamic similarity was maintained using dimensionless Modified Froude number. The splashing rate was found to increase with the increase of lance angle from the vertical and flow rate. The critical depth of penetration as well as the impact velocity for the onset of splashing was found to decreases with the increase of lance angle from the vertical. The dimensionless Blowing number (NB), which is a measure of droplets generation rate, was found to increase with lance angle if the axial lance height is kept constant. It was concluded that the Blowing number theory fails if the cavity operates in penetrating mode. The cavity surface area was predicted to be the most important factor in the generation of droplets at lower lance height but after a certain lance height the jet momentum becomes the dominant factor. Finally, effort was made to quantify the droplet generation rate from CFD model and validate against the present experimental study. It was found that the modelling of splashing rate using the Eulerian approach requires a very fine mesh to capture the fine droplets which are computationally very intensive. The CFD models, developed in the present study, can be used to predict the jet behaviour and impact on the liquid melt inside an industrial furnace at different nozzle inlet conditions. It can also predict the cavity dimensions and oscillations with reasonable accuracy. Thus, the models can be used as predictive tools in the industry. The knowledge of impact velocity and cavity dimensions is important to model the droplet generation rate from the cavity. Modelling of droplet generation is still a significant challenge for CFD but the results from this study indicated that Blowing number theory cannot be used to quantify splashing in the penetrating mode and a further modification is required by including the cavity dimensions as a parameter.
History
Thesis type
Thesis (PhD)
Thesis note
A thesis submitted for the Degree of Doctor of Philosophy, Swinburne University of Technology, 2012.