Four main topics related to non-uniformity in aluminium electrolysis cells are treated in this paper. i) The alumina concentration can vary by 2 wt% throughout the bath, which is the root cause of the low voltage anode effect (LVAE). By using a statistical method, a general chart separating the fields for LVAE and high voltage anode effect (HVAE) could be derived. Both types of AE occur at higher average alumina concentration when the concentration is less uniform. ii) Anodic current distribution can be used as a rough measure of the standard deviation in the anode-cathode distance (ACD); e.g., 12 % standard deviation for the current distribution corresponds to 5 mm standard deviation for the ACD. It is suggested that variation in ACD is caused by perturbation of the metal pool surface each time anodes are replaced due to Lorentz forces and bath density variations. iii) A simple formula for the effect on local ACDs by local bath density variations due to non-uniform alumina concentration was derived. It was shown that the time for equalisation of individual ACDs by current dependent anode wear is longer than the time between anode shifts. This entails that the ACD always varies with time and position. iv) Non-uniform ACD may be detrimental to the current efficiency (CE) because gas bubbles and metal waves can contact each other underneath anodes with low ACD. The effect is more severe when the average ACD is low.

In industrial aluminum electrolysis cells, the distribution of alumina is often non-uniform. This can be attributed to several factors, such as bath movement, discrepancies in alumina feeding from different feeders, and variations in consumption rates, to name a few. Furthermore, the non-uniform distribution of alumina has some effect on the pseudo-resistance of the cell and its variability, which are typically used to control alumina concentration in the cell. However, this can lead to an unpredictable lack of alumina locally in the cell, resulting in localized anode effects that can transform into generalized anode effects. Therefore, to achieve independent utilization of the feeders for balancing the alumina distribution in the cell, the first step is to have a reliable estimate of the alumina distribution. To achieve this, distributed information, such as the anode currents, is required. This study uses a two-step approach for estimating local alumina concentration and anode-cathode distance (ACD). In the first step, an observer uses the cell pseudo-resistance to obtain an initial estimation of alumina concentration and ACD. In the second step, a second observer uses the individual anode currents to correct the initial estimations. This paper is focused on the design of the observers and the required signal processing to obtain reliable estimations. Finally, the proposed method is validated with data from a smelter.

This paper conducted an analysis on different characteristics in a cell operated by an aluminum smelter. The measurements showed that there is a gradient in the distribution of bath temperature, superheat, and alumina concentration in the cell. Aimed at these different characteristics, this paper carried out research and industrial experiments on local control of alumina concentration for 6 months. The results showed that the current efficiency of the test pot increased by 0.51 % compared to the reference pots, the DC power consumption decreased by 62 kWh/t Al, and the anode effect frequency decreased by 40 %. The experiments showed that the local control technology of alumina is very promising, and it will be extended to more pots in aluminum smelters.

Efficient dissolution of alumina is essential to maintain ideal conditions in an aluminum reduction cell. Too little alumina dissolved in the electrolyte provokes an anode effect, while a surplus favors the sludge formation under the metal pad. In the two cases, the energy and environmental performance of the electrolysis cell decreases rapidly, in opposition to the priorities of the aluminum industry. This work presents a model of the interaction of the main phenomena involved in the alumina dissolution process to answer the question: How do the chemistry, and bath properties influence the alumina dissolution kinetics? Simulation results are compared with those obtained from an extensive experimental parametric study to validate the behavior and understand the kinetics involved. From these results, the limiting factor to the raft dissolution is the diffusivity of the dissolved alumina in its pores. An increase in the excess aluminum fluoride and calcium fluoride in the bath increases the diffusivity of the dissolved alumina, thus increasing its dissolution rate. However, this benefit adversely affects the solubility of the alumina in the bath. Accordingly, the relationship between bath composition and raft dissolution kinetics is presented at different superheats to understand this delicate equilibrium. Subsequently, the differences in the dissolution conditions between the experimental setup and industrial cells are described. Finally, a pathway to reach an optimal composition for industrial cells is presented.

In the aluminium smelting process, the information on spatial variations in alumina concentration is crucial for online alumina feeding control. However, the harsh environment of smelting cells renders long-term continuous real-time measurement of alumina concentrations impractical. Model-based state observers can be utilised for real-time estimation of the spatially distributed alumina concentrations, but the challenge lies in addressing model uncertainties, such as those in bath flow velocities and anode-cathode distance. To address these challenges, an H∞ filter-based multilevel state observer is proposed for estimating spatial alumina concentrations. This method can estimate alumina concentrations within a reasonable range and is robust to uncertainties. The effectiveness of this method is validated through experimental studies.