SIM-AM 2023

Qualification of DED metal Additive Manufacturing by numerical simulation and AI-based monitoring

  • Chiumenti, Michele (CIMNE)
  • Molotnikov, Andrey (RMIT)
  • Herzog, Timothy (RMIT)
  • Moreira, Carlos (CIMNE)
  • Caicedo, Manuel (CIMNE)
  • Cervera, Miguel (CIMNE)

Please login to view abstract download link

The metal Additive Manufacturing is an appealing technology allowing for the fabrication of complex geometry with a design freedom impossible with more classical metal forming technologies. However, the integration of AM in the industrial manufacturing chain is still limited because of the associated difficulties in monitoring the printing process. The optimal printing parameters are those that guarantee a constant morphology of the melt-pool (e.g. its size and penetration in the substrate), a uniform layer thickness and a controlled microstructure. Unfortunately, when adopting high-energy power supply such as in DED or the WAAM technologies the evacuation of the heat is not as fast as in LPBF and consequently the heat is accumulated in the substrate. The temperature distribution in the AM component is not uniform and strongly depends on the deposition path (generally not optimized according to the actual temperature distribution and its evolution in time) and the heat losses through the external surface of the AM-part. This given, the process parameters, and in particular the power input, cannot be kept constant during the printing process. Without an active control of the power supply, the overheating of the melt-pool or the lack of fusion of the material are unavoidable. In both cases, the consequence is the lack of homogeneity and uniformity in the printed layers. Eventually, the surface roughness can cause the laser defocusing (lower performance) or even the collision between the printing head and the substrate. Our proposal consists of providing artificial intelligence (AI) to the DED manufacturing process, allowing for an active modulation and optimization of the process parameters according to the current temperature of the substrate and its capability to evacuate the energy delivery. Feeding on the same input as DED machines (G-Code format), our AM software can very faithfully reproduce the power delivery of the laser along its path, as well as predict the part cooling during repositioning pauses, waiting times, etc. The melt-pool temperature and morphology are used to continuously monitor and modulate the laser power supply. In this way, it is intended to add an active and automated control to AM manufacturing, qualifying this technology for its adoption and integration in the industrial manufacturing chain. This optimized printing strategy is proven through an experimental campaign to show the actual benefit of the high-fidelity simulation