Keywords: additive manufacturing, residual distortions, artificial intelligence
Summary:A critical element for the design, characterization, and certification of components produced by additive manufacturing (AM) processes is the ability to accurately and efficiently model the associated materials and processes. This is necessary for tailoring these processes to endow the final products with proper geometrical and functional features. Capturing these features in AM material requires to solve a multi-scale, transient, heat, and thermoplastic structural problem. To minimize the computational burden associated with solving these models, an artificial intelligence framework that does not require solving the thermo-mechanical problem, but is trained to it, is proposed. The framework has been applied to efficiently predict distortions due to residual stresses through the entire AM component.