Spatiotemporal evolution of defects in an additively manufactured alloy monitored through in-situ X-ray computed tomography

In-situ X-ray computed tomography (XCT) is often used to explore failure mechanisms of materials through the global information of all the detected defects or the typical ones. However, the methodologies proposed hitherto often fall short in capturing the evolution of individual defects and their spatiotemporal characteristics. Herein, we develop a method for tracking each individual defect from its initial state to the point of fracture, with the objective of studying the defect’s evolution as a function of the tensile strain. Our tracking approach represents defects as 3D point clouds and evaluates their morphological similarity with an emphasis on their local environmental similarity. This significantly improves the average tracking accuracy, compared to the existing methods. We further reconstruct the fracture topology tree using the tracking outcomes so that the original critical defect that leads to the ultimate fracture can be identified. Using this methodology, we analyze the spatiotemporal evolution of pores in 316L austenitic stainless steel fabricated using the laser powder bed fusion technique. The results reveal that the combination of morphology parameter ( M_defect ) and environmental interaction parameter ( F_env ) are efficient and simple tools to capture the evolving characteristics of defects during tension. They enable the accurate identification of the critical defect that leads to fracture, once the tensile strain is sufficient such that the undetected defects/imperfect boundaries become visible to XCT. Using them, a simple unsupervised learning method, which enables a reasonable prediction of the fracture-related defects at a given strain, is also proposed.