A computer model for relating powder density to composition, employing simulations of dense random packings of monosized and bimodal spherical particles

This paper describes a model for relating powder composition to resultant density, which employs a new approach for simulation of the random dense packing of spherical particles. The powders are modelled as assemblies of spherical particles which are either all of one size, or a mixture of two sizes (categorised as monosized and bimodal packings, respectively). Random dense packing is the usual type of arrangement for metal powders, consisting of spherical particles above around 100 mu m diameter (below this size friction can be dominant and vibration may be required for attainment of the powder 'tap density'). This type of packing can be regarded as the result of gravitational forces which act on the spheres, moving them so that gaps are filled, with the assembly reaching a state of minimum potential energy. This movement is simulated here by an algorithm employing a large number of randomly generated movements, which have the effect of moving the spheres towards a central position. The resulting simulation was found to predict a fractional density (fd) of 0.635 for a random dense packing of spheres, which can be compared to the widely accepted value of 0.637. A correlation factor of 0.989 was obtained when comparing simulated and experimental packing densities for bimodal mixtures of spherical bronze particles. The model has been developed for application to the powder metallurgy (PM) manufacturing process, where it can be used to generate recommendations for particle sizes and proportions by mass, required for production of metal powders with specified densities. (C) 1997 Elsevier Science S.A.