Failure Investigation of Reused Metal Powder in Additive Manufacturing: An FMEA Approach and the Role of AI-Based Support Tools

Reusing metal powder in additive manufacturing (AM) is beneficial since it is more sustainable and cost-effective. However, reusing the powder in consecutive cycles provides additional challenges due to degradation mechanisms affecting powder characteristics. This study collects, refine, and organizes fail-ures associated with powder reuse in SLM, considering articles from literature and expert interviews, for supporting Failure Mode and Effect Analysis (FMEA). Through a systematic approach, the definitions of the failures are analyzed from a semantical and ontological point of view in order to identify the limitations that can undermine the risk assessment in FMEA. Finally, for each identified limitation, AI-based solutions for improving failure investigation are suggested. The findings suggest that failure investigation in AM is limited by ambiguities in failure clas-sification, intricate interdependencies, unprecise definitions, and a lack of rigor in capturing nonlinear relationships, reducing the effectiveness of conventional analysis methods. Among AI tools, machine learning, computer vision, natural language processing, Bayesian networks, and real-time monitoring systems have proven effective in addressing these challenges.