Most industrial scale fluidized-bed reactors are cylindrical, and the cylindrical coordinate system is a natural choice for their CFD simulation. There are, however, subtle complexities associated with this choice when using the Two-Fluid Model. The center of the grid forms a computational "boundary" and requires special treatment. Conventionally, a free slip no-normal flow condition has been used which does not predict the hydrodynamics accurately even when predicted parameters are in good agreement with measurement. Another difficulty is posed by the extremely small cells near the grid center, especially when simulating small scale experiments. The presence of these small cells raises concerns over the applicability of the Two-Fluid Model and is known to result in slow simulation convergence. These issues are addressed in the present study and appropriate solutions are proposed including the centerline treatment and the use of a non-uniform grid. Finally, the study compares the Cartesian grid with the cylindrical grid for application to fluidization. It is shown that simulating a cylindrical bed using the cylindrical grid is not only more accurate but also more computationally efficient. The analysis presented along with the proven computational efficiency of the cylindrical grid is especially significant considering that modeling commercial scale reactors, with multiple solid phases and chemical reactions, not only will require accurate description of the fluidization process but will also be exceedingly expensive in terms of computational cost. (C) 2014 Elsevier B.V. All rights reserved.