Prediction of mechanical properties from microstructural features via machine learning approaches for additive manufactured various steel alloys

One of the critical challenges in additive manufacturing (AM) of metallic alloys is to achieve the desired properties such as mechanical properties, wear and corrosion resistance. Due to the many variables involved in additive manufacturing, achieving this goal by trial-and-error will be cost prohibitive and time consuming. The high upfront cost of determining the resultant mechanical properties of metal parts built by additive manufacturing significantly hinders widespread adoption in the industry. Therefore, predictive modeling capabilities of additive manufacturing processes are desired. The objective of this study is to establish data-driven models that can correlate the microstructural details to mechanical properties. Once established, such data-driven models can provide a quicker way of predicting mechanical properties based on given microstructure information and can reduce the cost of achieving desired mechanical properties by drastically reducing the number of experiments required. The major advantage of such an approach is that a data-driven model can be built based on the widely available but scattered data in the literature along with an additional small number of necessary experimental data and it can be used to predict mechanical properties for various manufacturing processes used and heat treatment conditions. Three case studies are presented here to show the capabilities of machine learning approaches to predicting mechanical properties such as yield strength, ultimate tensile strength and ductility. Various relevant microstructural features including grain morphology, cellular structures, and constituent phases are considered as needed for different materials so that highly accurate data-driven predictive models are established. These studies demonstrate that the machine learning approaches are a powerful method of predicting various mechanical properties with minimal lead time and cost.