Modelling the Wall Collision of Regular Non-Spherical Particles and Experimental Validation

Abstract

The importance of numerical calculations (CFD) for supporting the optimization and lay-out of industrial processes involving multiphase flows is continuously increasing. Numerous processes in powder technology involve wall-bounded gas-solid flows where wall collisions essentially affect the process performance. In modelling the particle wall-collision process in the frame of numerical computations the general assumption is that the particles are spherical. However, in most practical situations one is dealing with irregular non-spherical particles or particles with a certain shape, such as granulates or fibers. In the case of non-spherical particle-wall collisions in confined flows, additional parameters such as roughness, particle shape and orientation play an important role and may strongly affect the transport behavior. The change of linear and angular velocity of the particle depends on these parameters, specifically the orientation and the radius of impact of the particles. In order to improve the non-spherical particle-wall understanding and modeling, in this work regular non-spherical particle-wall collisions in three dimensions are studied experimentally and computationally. For that purpose, cylindrical particles impacting a smooth wall at different angles are used. Single particle motion is tracked in space solving for both the translational and the rotational motion whereby the orientation of the non-spherical particle is obtained through the Euler angles and the Euler parameters. Once the particle touches the wall, the change of translational and angular velocity is determined by the non-spherical particle wall collision model. Experiments are made by shooting cylindrical non-spherical particles against a smooth plane wall at various impact angles and velocities. The collision event is recorded by two perpendicular arranged high-speed cameras. The experimental velocities obtained are used for validating the model.