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

Résumé

Carbon anode is one of the key components in the production of primary aluminum. The important desired properties of the anodes are high density, low electrical resistivity, low air and CO2 reactivities, and high mechanical strength. The anodes consist of pitch as binder and dry aggregate (coke, butts, and recycled anodes) as filler material. Granulometry of the dry aggregate is one of the key parameters that control the anode properties. In this article, a multilayer feed forward artificial neural network with backpropagation training has been used to correlate the dry aggregate granulometry with its bulk density. Experimental bulk density values of different size fractions of the dry aggregate were used for the training of the neural network. The model helps understand and predict the effect of different dry aggregate size fractions on its bulk density. This article presents the model and the results of the study.