The paper presents a three-dimensional (3-D), time-dependent Euler-Lagrange multiphase approach for high-fidelity numerical simulation of strongly swirling, turbulent, heavy dust-laden flows within large-sized cyclone separators, as components of the state-of-art suspension preheaters (SPH) of cement kilns. The case study evaluates the predictive performance of the coupled hybrid 3-D computational fluid dynamics–dense discrete phase model (CFD-DDPM) approach implemented into the commercial general purpose code ANSYS-Fluent R16.2, when applied to industrial cyclone collectors used to separate particles from gaseous streams. The gas (flue gases) flow is addressed numerically by using the traditional CFD methods to solve finite volume unsteady Reynolds-averaged Navier-Stokes (FV-URANS) equations. The multiphase turbulence is modeled by using an option of Reynolds stress model (RSM), namely dispersed turbulence model. The motion of the discrete (granular) phase is captured by DDPM methodology. The twin cyclones of SPH top-most stage have been analyzed extensively both for the overall pressure drop and global collection efficiency, and for the very complex multiphase flow patterns established inside this equipment. The numerical simulation results have been verified and partially validated against an available set of typical industrial measurements collected during a heat and mass balance (H&MB) of the cement kiln.