Simulation modeling for optimal control of additive manufacturing processes

This paper proposes computational models of the direct energy deposition and powder bed fusion processes developed for process control applications. Both models are built upon a regression metamodel of heat transfer beneath the laser beam, to which an auxiliary thermal model is added to account for residual heat in track-to-track interactions. Both models are coupled by taking temperatures predicted with the auxiliary model and incorporating them as initial conditions for metamodel predictions of future laser scans. The synergy of the metamodel and the auxiliary model creates a high-fidelity model, which is used to generate training data for a model-free optimal controller. Simulation results prove the capability of the proposed optimal controller to adjust scan speed to control temperature when accounting for track-to-track interactions.