A thermochemical dynamic model of a Top Submerged Lance furnace: Experimental validation with focus on minor elements for the Circular Economy

The trend towards a more circular economy presents a unique challenge for the pyrometallurgical engineer. Secondary feeds bring complexity to the smelter in the form of non-traditional chemistries and minor elements. Models of furnaces will play an important role in this paradigm. Models should be able to predict operations in dynamic systems that do not always operate at equilibrium.The development of a top submerged lance (TSL) model was the subject of this study because the TSL has proven to be capable of treating secondary materials.The furnace consists of a vertical cylindrical vessel, containing molten slag and bullion at the bottom. A lance enters through the roof and the tip is submerged in the slag, into which gas and fuel are blown. Secondary or primary feeds, fluxes and reductants can be fed to the furnace. The reactions and interplay between the liquid phases, gas and added reductants set the temperatures and partial oxygen pressures in the furnace. The Connected Local Equilibria (CLE) method was followed to model the furnace. This approach offers the benefit that speciation can be modelled simultaneously for many elements from thermochemical databases. The methodology is to divide the furnace into several equilibrium volumes, based on expected material flows. With each time step, equilibrium in each volume is calculated by Gibbs free energy minimization. Material is then exchanged between volumes according to expected flows. To validate the method, small scale crucible experiments were carried out. Molten lead-containing ferric calcium silicate slags (PbO-FCS slag, also containing GeO2, TeO2 and SnO2 in concentrations ˂ 1 wt%) were reduced under controlled CO/CO2 atmospheres to produce lead bullion. The kinetics of the process were measured. Similar experiments were carried out with a copper-containing ferric calcium silicate system. The CLE method was applied to simulate the data, using HSC Sim software. The crucible was divided into four equilibrium v