DEM modelling of powder flow and powder filling during die compaction

Powder compaction is a commonly used process in many industries such as powder metallurgy, pharmaceuticals and catalysts manufacturing. A homogeneous packing of powder bed in the die is an important prerequisite for consistent final products. Heterogeneities of packing such as pores, agglomerates or segregations can be induced during the die filling step. During die filling a powder is delivered from a feeding container (e.g. shoe) into a confined die and going from a quasi-static to dynamic flow. Both powder nature (particle size distribution, shape, cohesion) and die system design (atmosphere, die shape, shoe speed etc.) can influence the filling performance. Freeman and Xu[1] showed recently that the die filling performance can be predicted by measures of powder flow properties in an FT 4 powder rheometer (Freeman Technology, Malvern, UK; hereby referred to as FT4). Recently, the Freeman FT4 powder rheometer has become an increasingly popular device for studying the flowability of particulate materials by measuring the flow energy. Measuring flow energy consists of a dynamic test regime, in which the flow resistance encountered for an impeller moving within a powder bed is assessed. The FT4 has potential for use as a quantitative support tool for process design or a possible calibration tool for DEM simulations[2]. The downward dynamic test on the FT4, which provides a measure of basic flow energy (BFE), produces shear whilst compacting the sample and was found to be highly differentiating, particularly for non-cohesive powders. Meanwhile, the upward dynamic test, which provides a measure of specific energy (SE), does not compact the sample and correlated well with shear cell results. In this study, both the BSE and SE measurements were conducted in the FT4 powder rheometer using different types of alumina powders; BSE and SE tests were modelled correspondingly using a discrete element method. Particle properties such static and rolling friction coefficients, cohesion energy density, density were varied until the modelled BSE and SE values match the experimental values. Finally, die filling was simulated using these fitted values and the calculated filling behaviour compared with experimental data.