Characterization of Powder Flowability in Thermal Spraying using Machine Learning

The increasing trend in the use of fine powders in thermal spraying has been observed more frequently in recent years. The flowability of powder materials significantly influences their feeding behavior. This study aims to identify key characteristics for powder conveyability. A wide range of commercially available powder materials was considered, covering particle size distribution focusing on finer powder fractions, morphology, and chemical composition. Powder flowability is determined using a Hall Flowmeter, Hausner ratio, angle of repose, and avalanche angle measurement, with each method providing characteristic values as an indication of flowability. To correlate these values with the flowability of the powder materials, powder morphology is measured. Optical particle analyzer is utilized to measure particle size distribution, aspect ratio, sphericity, and convexity for geometric classification. Machine learning approach will be employed to analyze the measurement results and to rank the importance of various indicators in predicting powder flowability. The statistical analysis allows to determine the significance of these indicators of powder flowability. This machine learning model enables the analysis of correlations between various test methods for measuring powder flowability and powder morphology as well as particle size distribution.