(44b) Flowability of Dry Spent Coffee Ground (SCG) Powders

Coffee, which has an annual production of 9 million tons and a revenue of around US$ 27 trillion is the world's second most-valued traded commodity, after petroleum. Coffee powders can be forwarded to soluble coffee industries (SCI) or for direct use in beverage preparation contributing to produce a significant amount of spent coffee ground (SCG) powders. SCG is the most abundant coffee by-product with proven applications in composting, animal feeding, recovering of antioxidant components and as a biomass source for production of renewable energy by gasification, combustion or pyrolysis. In 2016, the SCI alone generated around 2.5 million ton of SCG readily available for industrial processing. In the most of the processes, SCG is submitted to feeding, storage, compaction and transporting operations whose success is highly dependent on the powders' bulk properties and flowability. Because the SCG powders come from quite different and uncontrolled sources, they usually present variable mean sizes, multimodal particle size distributions (PSD) and wide size span, making the prediction of their flowability behavior a challenging task. Using non-mechanical valves to feed dry SCG powders in a pilot scale circulating fluidized bed reactor is currently under investigation in our research group, motivated by its appealing potential for renewable energy production. During cold flow tests in a non-mechanical L-valve, it was observed that feeding stability and mass fluxes depend significantly on the mean size and flowability features of the SCG samples. This study was proposed aimed at evaluating the bulk density and flowability indexes of SCG powders of different PSDs and in a broad range of sizes, as well as testing whether the Linear-Mixture Packing model might be applied to predict the bulk properties of the mixtures. The powders were obtained after brewing coffee with water at 100ºC and oven drying the SCG at 105±2ºC for 24 h. The particle size analysis of the dried SCG original sample (X_d.b.=0.06) showed a trimodal PSD, with sizes ranging from 75 to 800 μm. The original sample was sieved into three base samples with restricted sizes, denominated A (d_A50=550 μm), B (d_B50=400 μm) and C (d_C50=225 μm), each one with a mean size representative of the major fractions identified in the original sample. Powder mixtures were produced by combining mass fractions (y) of 20, 40, 60 and 80% of the base samples, resulting in 12 binary and 6 ternary mixtures. Loose (ρ_lb) and tapped bulk densities (ρ_tb) were measured respectively, for 0 and 1250 taps, according to the QAS/11450 standard procedure, and the Hausner ratio (HR) was calculated for each sample. Loose and tapped bed porosities (ε_lt and ε_bt) were calculated using the true density of SCG (ρ_p=1315±4 kg/m³) measured using helium. The tests were carried out in triplicate for the base powders, binary and ternary mixtures, therefore a total of 63 samples were analyzed. Considering first the base samples and based on the HR values, the flowability of the C powder was categorized as very poor, while the A and B powders were classified as having a fair flowability. The loose and tapped bulk densities of the A and B powders were quite similar (ρ_lb=390±20 and 380±20 kg/m³;ρ_tb=467±9 and 460±10 kg/m³ for A and B, respectively), as these two powders have close mean diameters. The loose and tapped bulk densities of the C powder (ρ_lb=262±2 andρ_tb=393±7 kg/m³) were lower than those of the A and B powders, which is probably due to the high intensity of van der Waals forces and aggregates formation which is expected for this smaller size powder. According to the literature, at a size ratio d_C50/d_A50 >0.154, as in this case, the packing of binary and ternary powders occurs mainly by the occupation mechanism, in which the specific volumes are rearranged throughout packing, as the smaller particles fill in the voids. Packing ruled by the occupation mechanism is controlled by all the components in the mixture. For the mixtures tested here, both loose and tapped bulk densities were limited by the values of densities observed for the base samples, a behavior that agrees with that reported in the literature. The flowability of mixtures was most affected by the mass fraction of the small particle sample (C), and depending on the amount of C, different flowability characteristics were determined. For powders with y_c=0, the flowability was fair; for y_c=0.2 the flowability was passable; in the interval 0.4≤ y_c ≤ 0.6 the flowability was poor and for y_c ≥0.8 the flowability was very poor. The bed porosities were presented in ternary composition diagrams, with 0.70≤ ε_lt ≤0.80 and 0.64≤ ε_bt ≤0.70. The regions of higher porosity occurred for mixtures with larger mass fractions of C. The Linear-Mixture Packing model was used to estimate the loose and tapped porosities and agreed well with experimental data, with deviations less than 5%. Therefore, this model can be used to predict bulk properties of dry SCG samples, once the powder size composition is known.

ACKNOWLEDGMENTS

The authors are thankful for the financial support from São Paulo Research Foundation (FAPESP - process number 2016/25946-2).