We performed an experimental and numerical investigation of pneumatic conveying of cohesive dairy powder. The experiments with fat-filled milk powder (FFMP) fines with an average particle size of 94 μm were carried out in a 2-inch diameter stainless steel pipe consisting of two 2.5 m horizontal sections connected to a 0.65 m vertical section by two bends of 0.4 m radius each. In addition to measurements of pressure drop and powder deposition, an optical technique was used to measure the dynamics (probability densities) of local particle volume fractions as a function of operating conditions. Numerical simulations were performed with a commercial discrete element modelling (DEM) software, EDEM®, coupled with the computational fluid dynamics (CFD) software, FLUENT®. The simulation results in terms of pressure drops and particle volume fractions were compared with the experimental data. A very satisfactory agreement was found. At low gas velocities, cohesive dairy powders easily re-agglomerate after the second 90° bend and then deposit at the bottom of the horizontal pipe. At higher gas velocities, results show intermittent dispersion of particles and less particle deposition is observed even at higher loading ratio.@en

The development of a novel numerical model for droplet drying is the topic of this paper. The three main stages of droplet drying are distinguished, viz. unhindered evaporation of a ’wet’ particle (the droplet), restricted drying at a falling rate due to the formation of a crust around a wet core, and inert heating of the dry porous particle. Each stage is mathematically detailed to replicate all phenomena occurring throughout the drying process. The focus, however, is on the falling rate drying regime which is described in terms of Stefan diffusion of water vapour through the pores of a thickening crust. To this end, the model needs the material properties. This permits the droplet characteristics to be determined by composition rather than through single-droplet drying experiments. Finally, the model is validated against five of such experiments from literature using skim milk. Good agreement is found at each comparative case for the particle mass and temperature throughout the various drying regimes providing that for good reasons in three cases a lower drying air temperature is applied than reported for the experiments. The model is capable of predicting the entire drying process at low computational cost and without requiring empirical input.@en