Evaluation of a modified void descriptor function to uniquely characterize pore networks and predict fracture-related properties in additively manufactured metals

Abstract

Subtle differences among additive manufacturing (AM) processing parameters lead to variations in pore networks and complicate the prediction of void-sensitive mechanical behaviors, including location of fracture. The current work expands upon a recently developed pore metric, the void descriptor function (VDF), by accounting for interactions among neighboring pores and stress concentrations induced by non-spherical pores or voids. The modified VDF is evaluated against 120 computationally generated fracture simulations and six physical tensile specimens of as-built laser powder bed fused IN718. The latter set of experiments, which include X-ray computed tomography measurements before and after deformation, enables evaluation against pore populations that are representative of defects commonly observed in AM metals. The modified VDF accurately predicts fracture location (within ±5% tolerance) for 94 out of 120 simulated specimens, representing 3.3%, 62.1%, and 59.3% increases in the number of accurate predictions in comparison to predictions based on the original VDF, the location of maximum cross-sectional area reduction, and the largest-pore location, respectively. In the experimental data set, the modified VDF accurately predicts the location of fracture in five out of six specimens compared to only two out of six using the original VDF, maximum cross-sectional area reduction, or largest-pore location. Also, the maximum value of the modified VDF was found to be more highly correlated than fraction porosity, pore size, reduced-cross section area, and total number of pores to the ultimate tensile strength, elongation to failure, and toughness modulus, suggesting that the modified VDF presented in this work could serve as a promising metric to assist with characterizing unique pore networks and predicting fracture-related properties in AM components.