Évènements

05 juil
05/07/2019 14:00

Sciences & Société

Soutenance de thèse : Shuai CHEN

Investigation of FEM numerical simulation for the process of metal additive manufacturing in macro scale

Doctorant : Shuai CHEN

Laboratoire INSA : LaMcoS
Ecole doctorale : EDA162 : Mécanique, Energétique, Génie Civil, Acoustique de Lyon

Additive manufacturing (AM) has become a new option for the fabrication of metallic part. However, there are still some limitations for this application, especially the unfavourable final shape and undesired macroscopic proprieties of parts built in AM systems. The distortion or crack due to the residual stress leads usually to severe problems. In an AM system, the final quality of a metallic part depends on many process parameters, which are normally optimized by a series of experiments. In order to reduce time consumption and financial expense of AM experiments, the simulation for AM process, especially macroscopic simulation, is a prospective alternative for metal additive manufacturing research. In this thesis, we first study the pre-processing of AM simulation on Finite Element Method. The AM process is a multi-physics problem of coupled fields. For layer level, the reconstruction of 3D model is conducted from the scan path file of AM machine, based on the inverse manipulation of offsetting-clipping algorithm. For part level, the 3D model from CAD is reconstructed into a voxel-based mesh, which is convenient for a part with complex geometry. These pre-processing work of different levels indicate the methodology for simulating AM process and acquiring macroscopic proprieties of a part made by AM technology. Two post-processing tasks are also studied, aiming at the relation between AM process and part’s final quality. In the first task, a PID controller for power-temperature feedback loop is integrated in AM process simulation and the PID auto-tuning is numerically investigated. In the second task, dataset of heating parameters and residual stress are generated by AM simulation. The correlation between them is studied by using some regression algorithms, such as artificial neural network. Both of the two tasks show the important role of AM macroscopic process simulation.

Informations complémentaires

  • Amphithéâtre Clémence-Augustine Royer, Bâtiment Jacqueline Ferrand (Villeurbanne)