| Abstract |
Metal Additive Manufacturing (AM) is a fabrication technology with enormous advantages in terms of design and manufacturing freedom. The energy and material savings are substantial thanks to the possibility of manufacturing much lighter components with an optimized shape which can also include specific functionalities beyond the reach of the more classical technologies (e.g. forging, casting, etc.).
Direct Energy Deposition (DED) is the most appropriate technology for the fabrication of large metal components suitable for the mechanical, aeronautical, or automotive industries, among others. Unfortunately, the complexity of the AM machine programming necessary for the optimal layer-by- layer deposition makes its use still limited in the industrial practice. Currently, only the expertise of the technicians operating these machines and costly trial-and-error processes allow for finding the optimal process window as a function of the material, the geometrical complexity of the components, the heat accumulation r the actual residual stresses induced by the process. As an alternative, the adoption of Artificial Intelligence (AI) is proposed to provide a real-time monitoring and control of the process parameters (e.g. power supply, head speed, material feed flow, etc.), thereby ensuring the optimal process windows for the AM process, disregarding the complexity of the printing job. The AI is based on the deployment of an Artificial Neural Network (ANN) and its corresponding machine learning (ML) algorithm, fed by both in-situ monitoring and the predictive capabilities of software platform available for the numerical simulation of the DED process. The simulation software together with the AI monitoring conforms the Digital Twin of the DED technology.
The objective of this proposal pretends to this technology to the industrial sector willing to adopt the metal AM into their manufacturing chain, increasing production flexibility, allowing for repairing operations, thus minimizing costs and both energy and material waste. |
