Modelling structural mechanics and dynamics is a key asset for automotive design. At the level of the full vehicle, this is especially time consuming and represents a bottleneck in the design process. Moreover, the models are subject to uncertainty and structural instabilities leading to different collapse mechanisms. When confronted with Uncertainty Quantification that requires a high number of queries to the deterministic model, the computational cost increases dramatically. Thus, in the framework of analyzing and predicting the uncertainty of the modelling response of complex structures subject to extreme loads, it is crucial to develop efficient Reduced-Order Models (ROM).
This project aims at exploring the concept of sub-structuring for automotive simulations, dividing the vehicle in different modules to be coupled, and representing each of them with the more suitable nonintrusive ROM. Moreover, the efficiency of the coupling is strongly dependent of the description of the interfaces and the possible dimensionality reduction of their description. Thus, the ROMs are expected to be devised for each of the modules but also to their interfaces. The ROM strategies for the modules are expected to be nonintrusive because they must be trained with the commercial codes validated for the industrial practice.
The candidate integrating this project will benefit of:
- Integration within the research group of an automobile industry leader (SEAT-CUPRA).
- Develop proficiency in Machine Learning algorithms and their application in engineering analysis.
- Drive innovation in the application of Machine Learning to CAE domain, unlocking new possibilities for automotive safety, structural analysis, and performance optimization.
The deadline for registration to the offer ends on October 22nd, 2024 at 12 noon.
Més posts de Recerca
Cap comentari:
Publica un comentari a l'entrada