Morphology characterisation of additive manufacturing powder through automated image analysis algorithms

Abstract

The quality of parts produced by powder-based additive manufacturing processes are strongly influenced by the feedstock powder properties. In practice, powder-based additive manufacturing users analyse powder morphology characteristics to maintain powder quality. In this thesis, a method of mounting powder and using standard metallurgical preparation techniques in combination with image analysis is investigated for characterising powder morphology characteristics. Ti-6Al-4V powders from different powder production methods, as well as from different stages in a powder reuse regime, are analysed. Results from the proposed method are compared to benchmark data from a commercially available powder testing facility. Powder particles are cross-sectioned during the preparation process, therefore, stereological corrections are required to estimate the particle size distribution of the powder. Three stereological correction techniques are investigated; the Scheil-Schwartz-Saltykov method, the Goldsmith-Cruz-Orive method, and a Finite Difference Method. The stereologically corrected particle size distributions provide improved estimations of the particle size distribution, but no correction produces a particle size distribution equal to that of the benchmark data. The Finite Difference method is recommended with cumulative mean absolute errors across the powders investigated rangingfrom2.4%to7.5%, compared to cumulative mean absolute errors ranging from 2.8% to 8.1% for the Goldsmith-Cruz-Orive method, and from3.5% to 8.7% for the Scheil-Schwartz-Saltykov method. Three measures of particle shape are investigated: circularity, convexity and solidity. Differences arise between the results obtained from analysis of the polymer-mounted powder and those obtained from loose powder analysis for each particle shape measure, but a good degree of error overlap occurs in all cases. It is demonstrated that the presence of agglomerated or irregular particles may be determined from particle shape measures obtained from analysis of the polymer-mounted powders. Overall, the proposed method is shown to be effective for estimating particle size and shape properties of additive manufacturing powders.

Date of Award	Jun 2025
Original language	English
Sponsors	Department for the Economy
Supervisor	Shaun McFadden (Supervisor), Sagar Nikam (Supervisor) & Emmett Kerr (Supervisor)