Additive Manufacturing (AM) enables the production of functional and complex parts in a resource-efficient way, only applying the material to the desired location. Especially Laser-


based-Directed Energy Deposition (DED-LB) enables an excellent trade-off between production time and part complexity. However, this metal AM technology is currently lacking


in process stability, which limits a further industrialization. To overcome this, various industrial and academic efforts are focusing on investigating the application of digital process representations, often referred to as Digital Twins (DT), to improve the stability of DED-LB. This contribution presents a complete framework for a digital twin of the DED-LB process consisting of three major pillars. At first, a modular digitization framework for DED-LB is presented that utilizes an edge-cloud computing methodology to fuse data streams from multiple sensors monitoring the laser-


induced melt pool during production. The second pillar incorporates a thermal simulation model to predict the essential melt pool characteristics before the actual printing process. As third, a physics-informed neural network approach is applied to substantially reduce the computational time and efforts of the simulation’s melt pool predictions. In summary, the


presentation showcases a path towards data-based process control of an industrial-grade laser-based manufacturing system by utilizing digitization, physics-based simulation, and


artificial intelligence.