A machine-learning model to predict tensile properties of Ti6Al4V parts prepared by laser powder bed fusion with hot isostatic pressing

In this work, a machine-learning model was established to predict the influence of post-treatment, hot isostatic pressing (HIP) parameters on the tensile properties of Ti6Al4V parts prepared by Laser Powder Bed Fusion (LPBF), also known as Selective Laser Melting (SLM). The database was established by collecting published re-ports on HIP treatment of LPBF Ti6Al4V from 2010 to 2019, employing the following online resources: Google Scholar, Science Direct, Research Gate, and Springer. Using the established model, it is possible to prescribe HIP parameters and predict properties after HIP for LPBF Ti6Al4V parts with high confidence. The model shows great accuracy in the prediction of Yield Strength (YS) and Ultimate Tensile Strength (UTS). Of the predictions, 87.5% were within a 5% error range for YS in all datasets, and this value was 100% for UTS. It was found that the YS and UTS are sensitive to the HIP parameters, including temperature and holding time. The initial YS and UTS of as-printed parts are weaker influencing factors for YS and UTS of final parts, as compared with the HIP pa-rameters. The model suggests that a HIP process with a holding temperature lower than 970 degrees C and a holding time of approximately 2 h is sufficient for LPBFed Ti6Al4V parts to achieve mechanical properties that comply with the ASTM standard of Ti6Al4V for surgical implant applications (Standard Specification for Wrought Titanium-6Aluminum-4Vanadium ELI (Extra Low Interstitial) Alloy for Surgical Implant Applications-UNS R56401-). However, the prediction of strain to failure showed lower accuracy. Only 62.1% of predicted strain results were within a 10% error range. The differences among the printing processes by multiple research groups may have resulted in large variations of structural defects and thus led to significant scattering of the elongation data. The model also shows that elongation after HIP has a higher dependence on the as-printed samples.