École Nationale Supérieure de
Techniques Avancées

Abstract

Numerical simulation of mechanical large strains is one of the most chalenging issues of research and developement projects at Michelin. The development of performant simulation tools requires continuous research of methods in order to find the best compromise between the code robustness and calculations costs. This was the main object of my graduation internship. Attached to the service team DTO/SIM, I participated in the 3 dimensions tire rolling study and simulation project, "R3D". Some 3D rolling calculations were already performed. However, these calculations were very expensive in terms of computation time and resources. During this internship, the mission entrusted to me is to improve the performance and the robustness of numerical methods used for tire rolling simulations. For this purpose, two different ways of code performance improvement are selected : The first one consists of the remeshing tools study, crucial for the dynamic calculation control strategy, in order to improve the quality of finite elements meshes and ensure dynamic adaptation. Therefore, some technics are introduced in the cutting tool. This tool is subsequently coupled with the mesh adaptation one. This work helped to optimize the convergence calculations chances and guarantee a better solution quality computed by the Michelin solver. As for the second part, it consists of an accelerating calculation method study. For this, I worked on numerical methods called "reduced basis methods". Aiming to define an approach based on the "reduced basis methods" and adapted to the 3D rolling simulations, various variants have been tested so that to set up a suitable strategy. This latter is then implemented in Michelin internal solver.