Design rules and in-situ quality monitoring of thin-wall features made using laser powder bed fusion

The goal of this work is to quantify the link between the design features (geometry), in-situ process sensor signatures,and build quality of parts made using laser powder bed fusion(LPBF) additive manufacturing (AM) process. This knowledgeis critical for establishing design rules for AM parts, and todetecting impending build failures using in-process sensor data.As a step towards this goal, the objectives of this work are twofold:1) Quantify the effect of the geometry and orientation on thebuild quality of thin-wall features. To explain further, thegeometry-related factor is the ratio of the length of a thinwall (l) to its thickness (l) defined as the aspect ratio(length-to-thickness ratio, l/l), and the angular orientation(l) of the part, which is defined as the angle of the part inthe X-Y plane relative to the re-coater blade of the LPBFmachine.2) Assess the thin-wall build quality by analyzing images ofthe part obtained at each layer from an in-situ optical camerausing a convolutional neural network.To realize these objectives, we designed a test part with a set ofthin-wall features (fins) with varying aspect ratio from Titaniumalloy (Ti-6Al-4V) material-the aspect ratio l/l of the thin-wallsranges from 36 to 183 (11 mm long (constant), and 0.06 mm to0.3 mm in thickness). These thin-wall test parts were built under three angular orientations of 0°, 60°, and 90°. Further, the partswere examined offline using X-ray computed tomography(XCT). Through the offline XCT data, the build quality of thethin-wall features in terms of their geometric integrity isquantified as a function of the aspect ratio and orientation angle,which suggests a set of design guidelines for building thin-wallstructures with LPBF. To monitor the quality of the thin-wall, inprocess images of the top surface of the powder bed wereacquired at each layer during the build process. The opticalimages are correlated with the post build quantitativemeasurements of the thin-wall through a deep learningconvolutional neural network (CNN). The statistical correlation(Pearson coefficient, l) between the offline XCT measured thinwall quality, and CNN predicted measurement ranges from 80%to 98%. Consequently, the impending poor quality of a thin-wallis captured from in-situ process data.