A Data-Driven, Non-Linear, Parameterized Reduced Order Model of Metal 3D Printing

Abstract: Directed energy deposition (DED) is a promising metal additive manufacturing technology capable of 3D printing metal parts with complex geometries at lower cost compared to traditional manufacturing. The technology is most effective when process parameters like laser scan speed and power are optimized for a particular geometry and alloy. To accelerate optimization, we apply a data-driven, parameterized, non-linear reduced-order model (ROM) called Gaussian Process Latent Space Dynamics Identification (GPLaSDI) to physics-based DED simulation data. With an appropriate choice of hyperparameters, GPLaSDI is an effective ROM for this application, with a worst-case error of about 8% and a speed-up of about 1,000,000x with respect to the corresponding physics-based data.