Detailed numerical studies were carried out using a hybrid Eulerian-Lagrangian model for heavy nickel particles dispersing in a turbulent gas flow. A second-moment closure model, based on curvilinear coordinates, was used for the prediction of the fluid flow field whereas an improved Lagrangian stochastic model was employed for the prediction of the particulate phase. The improved Lagrangian stochastic model has accounted for the turbulence inhomogeneity, turbulence anisotropy, and particle crossing-trajectories effect. In addition, the particle inertial effect that heavy particles may disperse more than fluid particles in the longtime limit was also taken into consideration based upon the recently published theoretic analysis. Numerical results were compared with available experimental measurements and with other numerical results for both the gas and particle phases. Various parametric studies, such as particle initial conditions, eddy time scale, eddy length scale, particle inertia and restitution coefficients were performed to understand how these parameters influenced numerical predictions. It was found that the present numerical results were generally in better agreement with the experimental measurements than those of the other modelers. (C) 2000 Elsevier Science Ltd. All rights reserved.