Constellation, le dépôt institutionnel de l'Université du Québec à Chicoutimi

Résumé

Carbon anodes are one of the key components of primary aluminum production. One of the desired properties of the anodes is low electrical resistivity. A proper understanding of the effect of different parameters on electrical resistivity can help produce better quality anodes. A model has been developed to predict the anode electrical resistivity. First, using the Kopelman model for the thermal conductivity of a composite material, the specific electrical resistivity was modeled for the solid part (coke/cokified pitch) assuming coke as the dispersed phase in the cokified pitch matrix. Then, the effects of the anode porosity, distribution of particles, and coke properties are incorporated into the model using an approach based on the work of Shimizu. A factor which is a function of particle size and other properties is introduced. This factor was estimated using the artificial neural network. Published data were used to validate the model.