Clément POT – Ph. D. mechanical engineering
Project Description
The main goal of this project is to predict the mechanical properties of extruded parts only by using the extrusion parameters. Currently, the industry manufacturers need to do plenty of tests by varying the extrusion parameters in order to achieve the requierements of their clients to reach their goals in terms of mechanical properties.
This way of doing is particularly expensive and new technologies would be a great opportunity to fulfill the technical specifications of extruded products, and Artificial Intelligence could be a very powerful tool to save money from these tests.
This project is the following of the Quentin Boyadjian and Pierre Maurel’s projects in extrusion to predict the texture and anisotropy of extruded parts. The aim of their project is to use simulations to predict texture and anisotropy of extruded parts, which should be compared in fall 2022 or winter 2023 to textures of real extruded products. This should validate their work on extrusion, and anisotropy will be the next subject of focusing : using texture to predict the values of the Lankford coefficient and obtain a strong numerical model.
Objectives
– Simulate the extrusion and the mechanical transformation occuring during the extrusion process
– Analyse the current microstructural, mechnical and texture properties of real extruded parts
– Link the mechanical properties with the extrusion parameters
– Design a deep neuronal network and train it to provide a strong model