The compaction of granular materials is a process encountered in a wide variety of systems – both naturally occurring and man-made. It is utilized for “green body” manufacturing by many industries from ceramics and metallurgy to the chemical and pharmaceutical ones. In fact, the currently most prevalent device for drug delivery – the solid tablet is produced through the process of powder compaction.

A clear understanding of the dependence of the mechanical properties of the solid product on both the characteristics of the starting materials and the system dynamics during the process itself is crucial for the effective design, control and scale-up of powder compaction. For this purpose, a quasicontinuum approach has been implemented to simulate the compaction of confined heterogeneous granular systems. The method focuses on the post-rearrangement stage of powder compaction where inter-particle voids have been filled-in and any further volume reduction can occur only via particle deformation. The developed tool is capable of simulating powder beds consisting of particles of different sizes, as well as material properties.

The quasicontinuum model is implemented using a finite element framework, in which the displacement of most constrained particles is computed based on the displacements of the nodes of an adaptive mesh. In this formulation inter-particle forces are resolved locally, while their relative displacement is represented using a constrained field. Equilibrium is enforced weakly using the principle of virtual displacement. The interactions between particles in different cells are handled in a way allowing for local operation and rendering symmetric tangent operators. History-dependent inter-particle bonding terms are used to simulate the mechanical strength of the finished compact. The initial configuration of a post-rearrangement powder bed is supplied by a static DEM model.

This method is particularly well suited for the dynamic computation of macroscopic variables, such as pressure and density, as well as the resolution of micro-level quantities, such as coordination number and loading paths. Since particle positions and inter-particle forces are computed explicitly, the model makes no assumptions regarding the homogeneity and isotropicity of the powder bed. It is, therefore, capable of providing local density distributions in the finished product. These can be used for the prediction of the product behavior during future loading, including the onset of fractures and instabilities caused by zones of higher and lower levels of compaction.