Two different approaches for modelling the particle-wall collisions, the frequently employed Johnson & Jackson model and the recently proposed Schneiderbauer model, were evaluated in a fluidized bed riser by comparing simulation results to experimental data over a range of fluidization velocities and solids fluxes. For the Johnson & Jackson model, it was shown that partial slip settings recommended for denser fluidization conditions (a specularity coefficient in the order of 0.1) failed to predict cluster formation at the walls at higher gas flow rates due to unrealistically large granular temperature generation in the near-wall regions. By reducing wall friction to settings approaching a free-slip condition (specularity coefficient in the order of 0,001), this problem is overcome by eliminating excessive granular temperature generation from over-predicted strain rates at the walls. However, this approach results in an overestimation of the downward velocity of the clusters at the wall in dense cases. Despite this shortcoming, predictions are remarkably accurate for most of the cases. The Schneiderbauer model, with model parameters close to recommended settings, performs similarly well for most of the cases, slightly under-predicting cluster formation at the walls in the dilute cases. Generally, it also predicts more realistic flow behaviour since it prevents dense clusters from falling rapidly at the walls. The Schneiderbauer wall friction model is therefore recommended for use in future studies of risers, since it is able to deliver reasonable results over a wider range of flow conditions than the Johnson and Jackson model, using a single set of friction parameters. Furthermore, it has the benefit of using experimentally measurable quantities as input. (C) 2016 Elsevier B.V. All rights reserved.