Numerical Analysis of Powder Flow using Computational Fluid Dynamics Coupled with Discrete Element Modelling

Powder materials exhibit complex fluid-like features when subject to flow processes. Bulk behaviour comes about from interactions between solid particles and between particles and fluids, e.g. air. The multitude and complexity of interactions render experimental research findings confined to the conditions examined. Advances in numerical modelling technologies enable to use numerical analysis as an alternative to experimentation. This thesis is concerned with the development of coupled Computation Fluid Dynamics-Discrete Element Modelling (CFD-DEM) equivalent to real pharmaceutical powders. Powders used in pharmaceutical formulations are typically fine, light, cohesive and loosely packed. Traditional DEM uses spherical particles which give solid volume fractions of around 0.6, whereas real pharmaceutical excipients have solid fractions below 0.3. Such differences become particularly important when considering flow processes involving air-powder interactions. The coupled CFD-DEM framework was developed from the open source LIGGGHTS-OpenFOAM platforms. An adhesive contact law was implemented and the key interactions laws were validated. Systematic numerical experiments were performed to reproduce particle packing characteristics, permeability and critical orifice diameter; with the purpose to identify a set of particle characteristics (particle size, shape, elastic properties, coefficient of friction between particles and adhesion between particles) that matches the behaviour of real powders reported in the literature. The particle shape (approximated using the clumped sphere method) and the surface energy were found of key influence. The calibrated CFD-DEM model was used to reproduce experimental measurements of mass flow rate of powders through an orifice under the influence of differential air pressure across the powder bed for which an empirical model exists. The numerical studies made it possible to relate the key empirical parameter of the model to particle properties and identified density and surface energy as key influences. This work allowed for generalisation of the empirical model and replacing experimentation performed on bulk powders with measurable particle characteristics.