Interphase drag coefficient is a critical input in two-fluid model (TFM) and discrete particle method (DPM). In this study, extensive TFM and DPM simulations were carried out to study of the hydrodynamics of gas-solid flows of monodisperse, rough particles in a dense fluidized bed using fourteen drag coefficient correlations available in literature, the simulation results were then compared to the direct numerical simulation (DNS) and experimental data of Tang et al. (2016). It was shown that (i) all TFM and DPM simulations result in a lower bed expansion rate than those of DNS and experiment, which indicates that the true interphase drag force is underestimated by all fourteen drag coefficient correlations; (ii) all DPM simulations can correctly capture, at least in a qualitative sense, the main hydrodynamic features obtained from DNS, but some TFM simulations with the exactly same drag coefficients show anomalous phenomena. This is not an indicator of the failure of the used drag coefficients but might be a sign of the deficiency of the used particle phase stress model in this specific situation; (iii) the time-averaged differential pressure drops of DPM and TFM simulations agree well with DNS results, but the characteristic frequency of pressure fluctuation found in DNS is overestimated by a factor of 2-3 by all DPM and TFM simulations; and (iv) the magnitude of local and global granular temperatures of TFM, DPM and DNS are all in a fair agreement, but the degree of the anisotropy of both local and global granular temperature found in DNS is systematically underestimated by DPM and TFM, especially, the local granular temperature in TFM simulations is inherently isotropic due to the fundamental assumption of the used kinetic theory. This study highlights the need of a better interphase drag coefficient and a better particle phase stress model for modeling dense gas-solid flow. (C) 2019 Elsevier Ltd. All rights reserved.