Prediction of separation performance of hydrocyclones by a PC-based model

The separation performance of hydrocyclone is affected by many variables, the values of which have to be selected carefully for achieving the optimal design and operation of such a separator. This selection can be done by either empirical or numerical model. The empirical model provides convenience for industrial use by simply correlating cyclone performance indicators with variables under specific conditions. The numerical approach is more general and often performs well in explaining the underlying fundamentals of hydrocyclone operations, though it can also predict the unit performance at the cost of much longer computation time. In this work, an empirical prediction model is developed based on the extensive numerical results acquired by Computational Fluid Dynamics (CFD). This model covers a broad range of conditions. In particular, for the first time, it systematically considers the effects of material density and feed solid concentration at different hydrocyclone geometries. The validity of the model is confirmed by the reasonable agreement between the predicted and measured results in terms of cut size, separation sharpness, pressure drop, and water split. The comparison of the model with others in predicting performance indicators and partition numbers against the measurements shows that the new model is more accurate over the ranges of the variables considered.