Computational Approaches for Modeling Fluid Bed Granulation Process

Understanding the fundamental multi-physics behind the fluid bed granulation process and its projections in macroscopic scale provide the essentials to fabricate granulate with desired material properties. A proper prediction of physicochemical properties and morphological characteristics of a continuous wet granulation process product, particularly in response to operation conditions, is contingent upon the knowledge of the contact pattern evolution and the microstructure formation under the effect of the fluidizing medium. In this regard, current study looks at unveiling correlation between the process conditions and the main mechanisms that determine the material properties of granulate during fluid bed granulation process.

A multi-component modeling approach was explored that enables us to understand the fundamentals of fluid bed granulation process through both statistical modeling and first-principle modeling tools. A supervised machine learning algorithm that can simulate the evolution of process variables during fluid bed granulation process was explored, including: particle diameter, particle size distribution and loss on drying. Model predictions agree well with experimental analysis. Currently the focus is on a numerical framework that integrates design of experiment and the findings of statistical models with multi-scale mechanistic models such as discrete element model and computational fluid dynamics models. Using such models in early and late stage development phases will enhance the development timelines and increase the process product understanding.