Good powder spreadability is essential in powder bed-based AM to prevent the deposition of irregular layers that will induce defects in the built part. Investigating the powder spreadability by directly spreading it in the printer, despite being the obvious way to do it, is usually not feasible practically as it requires a large amount of material to fill in the machine and is a waste of machine time. Previous studies have demonstrated the link between the Cohesive Index metric [1] of the GranuDrum (Granutools, Belgium) and the irregularity of the layer measured inside an SLM printer. The methodology has been recently published as an ISO/ASTM technical report [2]. The interface fluctuations are measured based on the analysis of optical images of the powder bed acquired with the camera device available in the printer. Therefore, the observed defects can change depending on the angle of the lighting used. The in-situ evaluation is thus interesting to get a measure of the global spreadability of the powder but does not allow to go deep in the analysis of the defects shape and size. Discrete Element Method (DEM) has gained a lot of interest in simulating particle-based materials. Investigating spreadability with DEM allows performing deep analysis of the powder layer defects. However, obtaining accurate results requires a precise calibration of the model parameters that is usually time consuming. We propose an approach using characterization results obtained with the GranuDrum to precisely calibrate the simulations with a digital twin of the GranuDrum system [3]. The calibrated virtual material is then used to investigate the influence of material parameters such as cohesive strength, particle size or shape, and recoater speed, on spreadability. Moreover, the results of the DEM simulations confirm the relation between the Cohesive Index and the spreadability that was previously observed experimentally.