Hybrid first-principles neural networks model for column flotation

A new model for phosphate column flotation is presented which for the first time relates the effects of operating variables such as frother concentrat ion on column performance. This is a hybrid model that combines a first-pri nciples model with artificial neural networks. The first-principles model i s obtained from material balances on both phosphate particles and gangue (u ndesired material containing mostly silica). First-order rates of net attac hment are assumed for both. Artificial neural networks relate the attachmen t rate constants to the operating variables. Experiments were conducted in a 6-in.-dia. (152-mm-dia.) laboratory column to provide data for neural net work training and model validation. The model successfully predicts the eff ects of frother concentration, particle size, air flow rate and bubble diam eter on grade and recovery.