Comparison of DEM Results and Lagrangian Experimental Data for the Flow and Mixing of Granules in a Rotating Drum Ebrahim Alizadeh, Francois Bertrand, and Jamal Chaouki Dept. of Chemical Engineering,  Ecole Polytechnique de Montr eal, C.P. 6079 succ. Centre-Ville, Montr eal, QC H3C 3A7, Canada DOI 10.1002/aic.14259 Published online November 15, 2013 in Wiley Online Library (wileyonlinelibrary.com) This work assesses the accuracy of the discrete element method (DEM) for the simulation of solids mixing using non- intrusive Lagrangian radioactive particle tracking data, and explains why it may provide physically sound results even when non-real particle properties are used. The simulation results concern the size segregation of polydisperse granules in a rotating drum operated in rolling mode. Given that the DEM is sensitive to simulation parameters, the granule properties were measured experimentally or extracted from the literature. Several flow phenomena are investigated numerically and experimentally, including the particle residence time, the radial segregation, and the radial variation of the axial dispersion coefficient. An analysis of the DEM model is then presented, with an emphasis on the Young’s modulus and friction coefficients. Finally, dimensionless motion equations and corresponding dimensionless numbers are derived to investigate the effect of simulation parameters on particle dynamics. V C 2013 American Institute of Chem- ical Engineers AIChE J, 60: 60–75, 2014 Keywords: discrete element method, radioactive particle tracking, Lagrangian method, solids mixing, size segregation, rotating drum, physical properties Introduction Granular materials are widely used in processes found for instance in the mineral, ceramic, cement, metallurgical, food, and pharmaceutical industries. A common device for han- dling granular materials is the rotating drum utilized for vari- ous purposes. Regardless of the geometrical simplicity of such drums, the flow of material within these devices is quite complex. Depending on the range of Froude numbers Fr 5 x 2 R D g  , the fill levels and the frictions coefficients between the particles and the drum, different types of trans- verse bed motions can be observed, 1 where x, R D , and g stand for the rotational speed, the drum radius, and the gravi- tational acceleration. The most critical regime for the pur- pose of mixing is rolling. 2 This type of motion is characterized by a uniform flow of particles on a free flow- ing layer (active layer) located at the top of the bed, while the particles in the large underneath layer (passive layer) are transported upwards by the solid body motion of the drum wall. It is well-known that in rolling mode, the bed has a flat surface inclined at a dynamic repose angle. 3–5 Understanding and controlling the granular flow and mix- ing behavior in a rotating drum (as well as in other tumbling blenders) are of paramount importance for many industries. Inadequate mixing may result in rejection of the finished product due to poor quality. Unfortunately, there is insuffi- cient knowledge concerning the mixing of granular materi- als. Therefore, further understanding of granular flow is required to better grasp the mixing mechanisms as well as to design more efficient installations. In the past 25 years, many attempts have been made to comprehend the flow, handling, and characterization of granular materials. In spite of such efforts, the mixing of granular solids is still not well-understood as fluid mixing, due to the complex dynamic behavior involved. To determine mixing time or measure blend uniformity, several measurement techniques have been proposed. Light induced fluorescence (LIF), near-infrared reflectance spec- troscopy (NIR), and effusivity are some of the techniques that are currently available. 6 Characterizing granular mixing is also possible via physical sampling (e.g., thief sampling), which, however, interferes with the matter and affects the measure itself. 7 To overcome the limitations of physical sampling and to measure granular flow dynamics, several nonintrusive methods have been developed. Unfortunately, the granular media is opaque, thereby limiting most of these measurement techniques (including optical and visual meth- ods) inside the granular assembly. Moreover, the available optical experimental methods are limited to visualizing the granular surface. Among these methods, particle image velocimetry (PIV) 8 and particle tracking velocimetry (PTV) 9 are primarily applied. Magnetic resonance imaging (MRI) is another technique, capable of visualizing the bulk of the granular bed. 10 However, difficulty in obtaining MRI signals from solid samples restricts this method, thereby rendering it generally inapplicable to a wide range of granules. 11 MRI cannot be used to examine the flow of granules in any type Correspondence concerning this article should be addressed to F. Bertrand at francois.bertrand@polymtl.ca and J. Chaouki at jamal.chaouki@polymtl.ca V C 2013 American Institute of Chemical Engineers 60 AIChE Journal January 2014 Vol. 60, No. 1