Data-Driven Modeling of Mechanical Properties for 17-4 PH Stainless Steel Built by Additive Manufacturing

This study examines the link between microstructure and mechanical properties of additively manufactured metal parts by developing a predictive model that can estimate properties such as ultimate tensile strength, yield strength, and elongation at fracture based upon microstructural data for 17-4 PH stainless steel. The main benefit of the approach presented is the generalizability, as necessary testing is further reduced in comparison with similar methods that generate full process–structure–property linkages. Data were collected from the available literature on AM-built 17-4 PH stainless steel, in-house tensile testing and imaging, and testing conducted by an AM company. After standardizing the image size and grain boundary extraction via image processing, the features such as grain size distributions and aspect ratios were extracted. By using artificial neural networks, relationships were established between grain size and shape features and corresponding mechanical properties, and subsequently, properties were predicted for novel samples to which the network had not previously been exposed. The model produced correlation coefficients of R2 = 0.957 for ultimate tensile strength, R2 = 0.939 for yield strength, and R2 = 0.931 for fracture elongation. These results demonstrate the efficacy of predictive models that focus upon microstructure–property relationships and highlight an opportunity for further exploration as predictive modeling of metal additive manufacturing continues to improve.