Abstract Scope |
Although the outlines of a defect-based process map for powder bed fusion additive manufacturing (AM) have been in place for some time, many of the details remain to be defined. At low speeds, keyholes become unstable and shed pores. Computer vision helps to quantify this transition in terms of keyhole depth and aspect ratio based on high speed synchrotron x-ray visualization. Lack of fusion porosity is dominated by (lack of) melt pool overlap which is seemingly straightforward but subtly dependent on melt pool shape. For the latter, direct visualization provides unique insight into the laser penetration such that the effective absorptivity varies with power density. At high speed and power, fluid flow behind the heat source often causes pile-ups that become frozen in place, resulting in severe variability in melt pool dimensions. Tomography or sectioning reveals defect structures and machine learning again provides new tools for analysis of defect structures. |