Reconciliation of empirical correlations and CFD results for hydrocyclone performance for application in process modelling

Process modelling packages used for purposes such as process design and control of mill circuits generally use empirical and semi-empirical correlations to determine the performance of each of the unit operations in the circuit. It is well known that each of these correlations is subject to uncertainty. This paper focuses on correlations that are presently used for hydrocyclones. Computational Fluid Dynamics (CFD) modelling holds promise for predicting performance of hydrocyclones, but such modelling is also subject to uncertainties, particularly when the solids concentration is significant; however, even though the predicted values may not be accurate, the dependence of predictions on operating conditions should be more reliable. This paper uses the results of CFD simulations to determine the dependence of the primary performance measures on geometry and operating conditions. Performance measures analysed are cut-size, pressure drop and water split, and the dependencies are compared with those given by the Nageswararao model and the Flintoff-Plitt model. The results point the way to a methodology in which the best aspects of experiment, CFD simulation and process modelling can be combined.