A Comparison of Initial 2D and 3D Computational Models to Physical Cold Models

A comparison of initial 2D and 3D computational models to physical cold models

for modelling oxygen lancing of pyrometallurgical furnace tap-holes

M.W. Erwee1,2, Q.G. Reynolds1 and J.H. Zietsman3

1: Pyrometallurgy Division, Mintek, Randburg, South Africa

2: Dept. Materials Science and Metallurgical Engineering, University of Pretoria, Pretoria, South Africa

Abstract

Most pyrometallurgical furnaces have one or more tap-holes, from which slag, alloy and / or matte is tapped. Although tap-holes differ in design, most are closed using tap-hole clay and opened by a combination of mechanical drilling and lancing with oxygen. Lancing with oxygen involves a thin steel tube, through which oxygen is transported, which is inserted into the tap-hole. The steel tube melts and ignites in the oxygen, providing energy to melt the tap-hole clay. During lancing, reaction of oxygen with the slag, alloy, matte, refractory, and gas is possible. All of these interactions (chemical and thermal) can influence the process, but more so the tap-hole assembly itself. This can lead to premature shutdown of the furnace.

To study the effect of lancing in a controlled manner, multiphase models are being developed. Both computational and physical models are used to study tap-hole flow, during lancing, through the entire project. The initial physical model is a simple assembly using water and air. A high-speed camera and wall pressure measurements are used to observe flow phenomena when injecting an air jet into a static bath. The computational model is developed with OpenFOAM. For this work, 2D and 3D computational models are compared to the physical cold model. The 2D and 3D models are benchmarked against one another and differences between 2D, 3D and the actual physical model are highlighted.

A typical example of the results between the 2D model and the physical cold model are shown in Figures 1 and 2, which has been taken from a recent paper by the same authors [1][i]. The example results show a comparison to pressure signals measured along the wall close to the lance (Figure 1) for the 2D model and physical model, along with a visual comparison between computational work and high-speed video work (Figure 2)

Changes to the computational model (extended to 3D) and the physical model (e.g. changes in how pressure signals are sampled) are to be shown in this poster.

Figure 1: Pressure signals for lance depth of 1 cm. Gas flow rate: = 7.5 L/min

Figure 2: Computational model results and high-speed video images taken for a lance depth of 1 cm and a gas flow rate of 7.5 L/min. Top row: The parameter α in the computational model (α = volume fraction of water; white matches the condition α=0, black matches α=1). Middle row: The computed pressure field (blue is 0 mbar, red is 14.72 mbar; the colour scale is linear). Bottom row: frames taken from the high-speed video footage.

[i]References

[1] Erwee, M.W., Reynolds, Q.G., Zietsman, J.H., Cromarty, R.D., Lexmond, A.S. “Towards computational modelling of multiphase flow in and around furnace tap-holes due to lancing with oxygen: an initial computational and cold model validation study”, Infacon XIV, Kiev, Ukraine, 1-4 June 2015.