Modelling of an anodic gas evacuation & current efficiency in Hall-Heroult cells

Abstract

Anodic gas evolution & evacuation influences many aspects of cell performance in aluminium electrolysis. Electrolyte flow induced by the release of anodic gas affects processes such as alumina dispersion/dissolution, mass transfer of dissolved species and their subsequent reaction, heat transfer and its related ledge melting/freezing, and the metal-bath interfacial behaviour. The anodic gas coverage of the anode causes additional cell resistance, and its periodic discharge causes fluctuation in the anode current, contributing to the cell dynamics. A major proportion of the current efficiency loss in aluminium electrolysis is due to the "back reaction" of the electrolytic products carbon dioxide and aluminum metal. This "back reaction" has generally been accepted as being controlled by the mass transfer of dissolved species, in which gas-driven electrolyte flow plays a role. Available measurements of current efficiency were reviewed and analysed, and several current efficiency models were critically examined. It was evident that these data are not always consistent. In particular, the current efficiency behaviour in relation to alumina concentration and anode-cathode distance remains uncertain. A reliable and complete current efficiency model is especially lacking, which takes into account the influence of all major cell variables, in particular, alumina concentration, current density, and anode-cathode distance. A physical analogous model to a reduction cell was constructed in this work to investigate the gas evacuation phenomena. Fluid dynamic quantities characterising anode gas release were measured, and some aspects of the bubbling mechanism were discussed. Visual observations indicate that eddies produced in the electrolyte phase result in a very complex metal-bath interfacial wave motion. Under certain conditions, these wave motions will cause sporadic breaking of wave crests, resulting in droplets of metal being dispersed in the electrolyte. A possible metal transport mechanism at the metal-bath interface involving interface deformation and droplet formation was suggested. The air-water-C6H5Cl analogue model used in this work is capable ofsimulating the overall mass transfer process involving bath convection as well as interfacial transport as found in reduction cells. Assessment of species concentration gradient within the "bath" allows the "back reaction" scheme to be determined. Measurements of mass transfer coefficients at the gas-bath interface and the subsequent measurement of species concentration also quantifies the metal-bath interfacial mass transfer coefficient. Dimensionless correlations were developed to fit the measured mass transfer coefficient data. The transition criterion of the mass transfer regime was also quantitatively defined, which includes the influence of major cell variables. Based on the proposed reaction scheme, in conjunction with the dimensionless correlations of the experimental mass transfer coefficients a mathematical model was formulated to predict the mass transfer driven current efficiency loss in reduction cells. The model has the advantage in using the experimental mass transfer coefficients obtained from a analogue model with dynamic similarity to the reduction cell, and thus involves less uncertainty. The model contains most of the major process variables including bath chemistry, bath cavity geometry, and the current density. With the original correlations used the model predicts the theoretical limit of current efficiency where only gas-driven flow mechanism is present. The model also simulates the current efficiency in an actual cell situation by adjusting the mass transfer coefficient to an appropriate level. Predictions of the current efficiency were compared with available industrial data, and the implications of agreement and differences involved are discussed. The results obtained do not support the suggestion that the non-linear behaviour of current efficiency with alumina concentration is a direct result of variation in bubble surface area. A possible mechanism with respect to the. drastic loss in current efficiency for a very low ACD is proposed, which involves the metal surface break-up and the subsequent metal droplets dispersion in the bath, as observed in the physical model. A possible control target for current efficiency based on the surface instability criteria was discussed. A numerical investigation of the boundary layer mass transfer process was made for an in-depth understanding of the mass transfer process at the metal-bath interface. The governing equations solved retain all the major transfer mechanisms which include molecular and eddy diffusion normal to the interface as well as convection associated with the boundary flow. The numerical model allows the interface to be wavy, and the metal pad circulation to be incorporated. With one adjustable constant in the eddy diffusivity model the numerical predictions gave reasonable agreement with the measurements. The model provides a tool for cell flow design (including MHD design) in relation to current efficiency. A bath flow simulation was performed with the PHOENICS programme to provide flow velocity and turbulent kinetic energy information. The simulation method used is not considered to be superior, but is expected to be reasonably accurate in providing flow information in the proximity of the metal-bath interface, as required in solving the boundary layer mass transfer equations.