Modelling of top submerged swirl and non-swirl gas injections into liquid baths

Experimental and numerical investigations were carried out on the fluid flow behaviour within cylindrical liquid baths agitated by a top submerged gas injection lance system. The effects of key hydrodynamic parameters, such as the swirl and non-swirl gas injection, the gas injection flow rate, the level of submergence and the depth and size of liquid bath, on the flow structure and the mixing process within the bath were investigated. Experimental and numerical results from the current study enabled to gain more knowledge and physical understanding of such system, and to determine the optimum operating conditions for the industrial systems. The Laser Doppler Anemometry (LDA) measurement technique was employed to determine the mean and turbulent flow structure in laboratory models of top submerged gas injection into liquid baths. Two cylindrical air-water models were constructed according to the geometric similarity criteria related to the real industrial smelting furnaces. The experimental operating conditions were then decided based on the Froude number dynamic similarity criterion. The three-dimensional flowfield structures of the liquid phase including the time mean velocity components and turbulence parameters were obtained with the aid of LDA under forty-two combinations of the operating conditions for both laboratory models. A digital video recording technique was also employed to study the plume characteristics due to the LDA limitation in obtaining the velocities in the plume region. Numerical simulations of the fluid flow structure within a liquid bath that is agitated by a top submerged gas injection system were carried out using the commercial Computational Fluid Dynamics (CFD) software package CFX4. A two-dimension gas-liquid transient numerical model was developed by using the Eulerian-Eulerian approach. The standard k-Ε model was employed to predict the turbulence structure of the liquid phase. The comparison of the numerical predictions with experimental data under various operating conditions showed a reasonable agreement on both mean and turbulent fluid flow characteristics