The defluidization of fine (mildly cohesive) glass particles with mean size 69 pm is studied by DEM-CFD simulations and experiments for quantitative model validation. Cohesion arising from van der Waals forces between particles is explicitly incorporated into DEM-CFD, along with the corresponding effects of surface roughness. Due to computational limitations, DEM-CFD simulations often use artificially soft particles and systems much smaller than experiments, making the direct comparison between simulation and experimental results questionable. This computational obstacle was recently overcome for coarse (non-cohesive) particles via the identification of a system-size-independent measurement, thereby allowing for the direct comparison between small-scale simulations and large-scale experiments (LaMarche et ai., 2015b). Here, we assess the applicability of the same measurement, namely the defluidization curve, to cohesive systems. Unlike non-cohesive systems, the simulation results exhibit a sensitivity to Young's modulus and (low) static bed heights. These observations are explained by the enhanced cohesive effect at lower Young's modulus and decreasing bed porosity with increasing static bed height. Nonetheless, by using the true material Young's modulus in the DEM-CFD simulations and sufficiently large static bed height, a system-size-independent defluidization measurement is achieved that compares well with experimental data from a much larger system. This ability to directly compare small-scale simulations and large-scale experiments via a system-size independent measurement (defluidization) allows for the quantitative validation of the particle-level models in DEM-CFD without resorting to particle-level measurements and/or unrealistically small experimental systems. (C) 2016 Elsevier Ltd. All rights reserved.