(62f) Modeling the Agglomeration Process in Fluidized Bed Granulation: a Simple Approach

Granulation, the size enlargement process finds application extensively in pharmaceutical, food, fertilizer and mineral industries. This can be achieved in many methods such as fluid bed granulation, high shear granulation, drum granulation, etc. In fluidized bed granulation, granulation fluid in the form of – solution, suspension, slurry, melt – is sprayed on the surface of the particles which are fluidized by air. While the particles enter the spraying zone, it captures granulation fluid on its surface. When these particles collide, it may stick with one another and granulation fluid forms bridge among particles. These bridges are later converted into solid bridges while receiving sufficient heat to evaporate the solvent present in it. Thus, granules are formed. The necessary steps for this granule formation – collision, bridge formation, coalescence, drying – depend on operating conditions and physical and chemical properties of both feed particle and granulating fluid. The prediction of particle size distribution (PSD) and average size is difficult on the basis of theoretical considerations alone since exact mechanism by which this size enlargement occurs is not fully established. Hence, there is a need of experimental investigation to model this process. Experiments are carried out investigate the effect of process parameters – fluidizing air velocity, binder flow rate – on granule growth and the observations are used to model the process.

In this work, modeling is focused to describe the granulation phenomena on the basis of experimental observations of wheat powder granulation in fluidized bed. The experimental observations of PSD changes with time indicate that the evolution kinetics is analogous to elementary reversible series reaction. In this modeling, only three classifications of the granule size are considered and their evolution with time is described by differential equations. The parameters used in the differential equations are estimated using optimization technique.