A meltpool prediction based scan strategy for powder bed fusion additive manufacturing

In this study a feedforward control method for laser powder bed fusion additive manufacturing is demonstrated. It minimizes the meltpool variation by updating the laser power based on a data-driven predictive meltpool model. A rectangular pattern is scanned multiple times on a customized LPBF testbed. The meltpool is monitored in situ by a high-speed camera, optically aligned with the heating laser. Constant laser power is applied for the first scan, and its meltpool images are used to train the model and adjust the laser power for the following scans. The meltpool images from these scans are compared, and a significant reduction in meltpool variation is achieved.