LAUSR

Abstract Flotation models based on parameters derived from sampling campaign data degrade over time if left unattended. This problem can be addressed by continuously regressing flotation rate constant distributions, using measurements that are available online. Different distributions were investigated, to determine how the modelling accuracy of concentrate flows from individual cells are affected by more complex distributions. Rate constant distributions were regressed using a comprehensive data-set based on a detailed sampling campaign, as well as combined concentrate-flows, to approximate results achievable using routinely available data on industrial plants. Two circuit configurations were used to show that parameters regressed for one configuration is also valid for another, hence being representative of feed characteristics.