Dynamique d'ecoulement et pelletisation dans un granulateur a rotor

The first part of this thesis studies the dense granular flow of a 2 and 4 mm particle blend inside a spheronizer. The use of the discrete element method (DEM), a particulate model that can simulate the particle motion based on Newton’s second law of motion, allows characterizing the particle flow behavior inside the equipment. The particle mixedness state is assessed with the help of mixing indexes that have been developed by Doucet et al. (2008) in order to quantify the segregation occurring inside the particle bed. This study shows that the fill level and the disc rotational speed have a significant impact on the segregation phenomena observed. For a disc speed varying between 20 and 100 rad/s, the particle bed takes the form of a torus within which two distinct segregation zones are observed. As the disc speed increases, the small particles tend to migrate from a zone located at the center of the torus toward another zone which is observed near the spheronizer wall. This transfer of small particles is confirmed by the coefficients of correlation used by the mixing index, which relate the particle size and spatial coordinates.;The second part of this thesis develops a new approach to incorporate and control interparticle forces homogeneously in the context of particle flow applications. This approach uses particles coated with a PEA/PMMA copolymer. When submitted to an increase of temperature above the copolymer glass transition state, the interparticle forces increase. A relationship between the interparticle forces created by the copolymer and the flow of coated particles is characterized with a surface force apparatus (SFA). This equipment shows that the cohesion forces increase linearly when the temperature in incremented from 10°C to 50°C. The interparticle forces obtained are in the same range as other common forces encountered frequently in granulation processes such as the capillary and the van der Waals forces. The flow behavior of the cohesive coated particles is applied to two different applications. The first application considers a dense particle flow normally encountered during a wet granulation with a modified spheronizer. The second application shows the possibility to mimic the particle flow behavior that would be obtained in high temperature fluidized beds but with the advantage of operating them near ambient conditions.;The third part of this work uses the polymer coating approach proposed in the second part of the thesis to characterize the flow behavior of cohesive particles inside a modified spheronizer. By controlling the level of intensity of interparticle forces with the increase of temperature, four different flow states are observed. The first state is characterized by a free-flowing behavior of the particles, which is observed near ambient temperature. The second flow state is associated with the appearance of agglomerates at the surface of the torus of particles. These agglomerates increase in size as the temperature is incremented within this state. The third flow state refers to the appearance of a secondary layer formed by agglomerated particles the volume of which changes periodically with respect to time. The fourth state is characterized by a solid mass motion of the particle bed which is produced following the complete agglomeration of the particles.;The fourth part of this thesis develops of a multiscale model based on an event-driven Monte- Carlo based population balance. It is used to simulate wet granulation in a modified spheronizer. The model takes into account the particle scale with three different granulation mechanisms, which are the wetting, the coalescence and the breakage of the particles. On the other hand, the granulator scale integrates the particle motion with a compartmental approach which divides the particle bed into different zones, each of which is associated with a granulation mechanism. The use of a continuous-time Markov chain allows representing the motion of the particles between the different zones. The properties of the Markov chain, which are the residence time in the different zones and the matrix containing the probabilities of transition between the zones, are built with the DEM simulation results of the particle flow in the spheronizer presented in the first article. Once the multiscale model is defined, it is compared to a conventional population balance that does not take into account particle motion. These two population balance models are then tested against granulation experiments with the modified spheronizer. The results show that for a low spray rate, the multiscale model improves results obtained with the conventional population balance without motion. On the other hand, the multiscale model and the conventional population balance give similar results when the spray rate is high. In this case, the granulation mechanisms overcome the effect of the particle flow pattern and the advantage of the proposed multiscale model is less apparent. (Abstract shortened by UMI.).