AI for abnormal and critical grain growth phenomena discremination and avoidance – Application to Nickel base superalloys

Annual gross salary: about 26k€

Mines ParisTech CEMEF, Sophia-Antipolis (06)

Personnes à contacter par le candidat

marc.bernacki@mines-paristech.fr

TÉLÉCHARGEZ L’OFFRE

Abnormal grain growth (AGG) and critical grain growth (CGG) are two well-known metallurgical mechanisms leading to overgrown grains in microstructures without or with stored energy, respectively. As such large grains can be detrimental to the fatigue resistance, it is a critical issue for numerous industrial applications. The understanding and full-field modeling of these mechanisms has greatly been improved in the last decade. A statement at the basis of this proposal is that the partners already have the capability of modeling in full-field context the main mechanisms that can lead to AGG or CGG during annealing.

However, from the industrial perspective, the complexity remains that it is generally impossible by looking the final experimental micrography of the microstructure to discriminate easily and confidently the elements involved in the occurrence of these phenomena and their chronology. This aspect explains one of the current difficulty of optimizing the obtained microstructures and why trial-and-error methods often remain the industry norm. Indeed, the causes of AGG or CGG could be numerous when considering the applied TM paths and their effect at the microstructure scale. Finally, the concept of the AI-for-AGG proposal will be to move from our simulations’ predictive nature to an intuitive understanding of final microstructures caracteristics presenting AGG or CGG or even to end up with processing maps exhibiting the windows where the risk of AGG or CGG exists.

This breakthrough objective will be based on our capability to build a massive database concerning available experimental data and numerical predictions of AGG/CGG taken into account all the possible causes and the use of neural network-type algorithms to develop new capabilities on AGG/CGG discremination and avoidance.