A Pragmatic Model for Alumina Feeding

In this paper, we demonstrate how we can develop a coarse-grained model for the alumina distribution in an aluminium reduction cell. The model is designed to have the potential of being a part of a control system or play the role of a Digital Twin. The modelling work is applying the concepts of Pragmatism in Industrial Modelling [5]. The task is to be able to dynamically keep track of where dissolved and undissolved alumina are inside the cell. The numerical grid is the coarsest possible, and special numerical techniques are applied to support fast simulations. The bath (electrolyte) flow is obtained from detailed CFD simulations and imported into the coarse-grained model. A method to deal with dispersion in such a coarse-grained model is developed. The physics of particulate alumina dissolution and the electrochemical consumption of dissolved alumina at the anodes are represented. A simple model for the current distribution through the anodes is applied. The model is typically running much faster than real time. In demonstration simulations, the model runs 50-500 times faster than real time. From these, it can be observed how alumina particles and dissolved alumina distribute in time and space. Regions where anode effects are expected to initiate can be observed as well as the impacts of changing the feeding pattern and positions for the alumina. The numerical approach is inspired by previous works [9, 10]. The possibilities and limitations of the approach are discussed.