Enhancing Part Quality Management Using a Holistic Data Fusion Framework in Metal Powder Bed Fusion Additive Manufacturing

Abstract

Abstract Metal powder bed fusion additive manufacturing (AM) processes have gained widespread adoption for the ability to produce complex geometries with high performance. However, a multitude of factors still affect the build process, which significantly impacts the adoption rate. This, in turn, leads to great challenges in achieving consistent and reliable part quality. To address this challenge, simulations and measurements have been progressively deployed to provide valuable insights into the quality of individual builds. This paper proposes an AM data fusion framework that combines data sources beyond a single-part, development cycle. Those sources include the aggregation of measurements from multiple builds and the outputs from their related models and simulations. Both can be used to support decision-makings that can improve part quality. The effectiveness of the holistic AM data fusion framework is illustrated through three use case scenarios: one that fuses process data from a single build, one that fusses data from a build and simulation, and one that fuses data from multiple builds. The case studies demonstrate that a data fusion framework can be applied to effectively detect over-melting scan strategies, monitor material melting conditions, and predict down-skin surface defects. Overall, the proposed method provides a practical solution for enhancing part quality management when individual data sources or models have intrinsic limitations.