The complex nature of multi-phase flow in hydrocyclones has led designers to rely on empirical equations for predicting the performance. These empirical relationships are derived from an analysis of experimental data and include the effect of operational and geometric variables. A number of classifying cyclone models have been developed over the past three decades (Plitt, 1976 and Nageswararao, 1978). The problem with empirical cyclone models is that they cannot be used outside the range of conditions under which they were developed. A better empirical cyclone model using mathematical structures based on fluid mechanics is highly desirable. In this paper simulations of various cyclones were conducted in FLUENT where the particle phases were simulated using the Mixture model. The turbulence was resolved using Large Eddy Simulation (LES) model. The simulations confirm that the tangential velocity is the key in separating the particles. Further multi-phase analysis shows the turbulence dispersion also affects the fine particle segregation. Also, the experimental and the CFD data indicate that particle separation inside the cyclone is also affected by the feed solids concentration and size distribution, primarily in terms of the slurry viscosity and the particle hindered settling. Mathematical correlations have been developed which include these effects and have been incorporated in an improved empirical hydrocyclone model.