Merging Machine Learning and TriBeam Tomography for 3D Defect Detection in an AM CoNi-Based Superalloy

Abstract

Energy efficiency is a strong driving force for development of 3D printing processes for geometrically complex high temperature superalloy components. However, defects in additively manufactured (AM) multicomponent materials are currently a significant barrier. Superalloys have diverse chemistries that can result in complex solidification paths, heat affected zones, and liquation cracking [1,2]. It is well understood that selective laser melting (SLM) induces steep thermal gradients during printing that impose large thermal strains on the material [3]. These strains, in addition to chemical segregation during solidification, lead to significant cracking during AM, limiting mechanical performance. In addition to cracks, defects such as pores, voids, and inclusion phases can form during manufacturing due to either print parameters or material chemistry [4].