Determination of a quantitative relationship between material properties, process settings and screw feeding behavior via multivariate data-analysis.

In this study, a quantitative relationship between material properties, process settings and screw feeding responses of a high-throughput feeder was established via multivariate models (PLS). Thirteen divergent powders were selected and characterized for 44 material property descriptors. During volumetric feeder trials, the maximum feed capacity (FCCmax), the relative standard deviation on the maximum feed capacity (RSDFCmax), the short term variability (STRSD) and feed capacity decay (FCdecay) were determined. The gravimetric feeder trials generated values for the mass flow rate variability (RSDLC), short term variability (STRSD) and refill responses (Vrefill and RSDrefill). The developed PLS models elucidated that the material properties and process settings were clearly correlated to the feeding behavior. The extended volumetric feeder trials pointed out that there was a significant influence of the chosen screw type and screw speed on the feeding process. Furthermore, the process could be optimized by reducing the feeding variability through the application of optimized mass flow filters, high frequency vibrations, independent agitator control and optimized top-up systems. Overall, the models could allow the prediction of the feeding performance for a wide range of materials based on the characterization of a subset of material properties greatly reducing the number of required feeding experiments.