Abstract Scope |
In the past few decades, additive manufacturing has evolved for the one-step fabrication of various complex, customized metallic components that cannot be easily and economically produced by other means. However, widespread applications and market penetration of such components are often hindered by the formation of common defects that affect part quality, reliability, and serviceability, and increase the cost. Reduction in surface defects which is essential for long fatigue life and high dimensional tolerance is currently achieved by post-process machining and grinding that add extra costs and is not viable for parts with internal channels. Here, for the first time, physics-informed machine learning, mechanistic modeling, and experimental data are combined to reduce the surface defects in additive manufacturing. By analyzing one hundred and sixty-six independent experimental data for six different alloys on the defect formation available in the disjointed, peer-reviewed literature, several important variables that reveal the physics behind the surface defect formation are identified. Values of those variables are computed using a mechanistic model and analyzed in a physics-informed machine learning to provide the hierarchical importance of the variables on defect formation. In addition, based on the results of the physics-informed machine learning, an easy-to-use, verifiable, quantitative equation is provided that can be used in real-time to predict and reduce surface defects before experiments. The proposed methodology can help in reducing common defects such as balling, cracking, lack of fusion, porosity, and rough surface, and solve other complex engineering problems beyond additive manufacturing. |