In industrial applications, fluidized bed reactors with complex geometry are often used to process non-spherical particles, e.g., the biomass utilization. It is well known that CFD-DEM (Computational Fluid Dynamics/Discrete Element Method) is a powerful tool for studying the fluidization. However, few researchers had ever investigated the fluidization of non-spherical particles in the complex geometry using CFD-DEM. In this study, a robust and efficient CFD-DEM algorithm is correspondingly proposed, in which the non-spherical particles are represented by rod-like particles and are described by the super-ellipsoid model. An approach is incorporated into the CFD-DEM model to detect the contact between the rod-like particles and the complex geometry. The CFD-DEM coupling method has also been modified for the coupling between DEM and CFD with the unstructured mesh. Subsequently, the efficiency of the coupling method of CFD-DEM is tested. To validate the prediction accuracy of the CFD-DEM model, the simulations of the fluidization of rod-like particles in a fluidized bed with a horizontal immersed tube that represents the complex geometry are then compared with the corresponding experiments. The results demonstrate that the proposed CFD-DEM model has a rather high efficiency and accuracy for modeling the fluidization of rod-like particles in a fluidized bed with complex geometry. (C) 2018 Elsevier B.V. All rights reserved.