ARCO Improvement for Seat

SEAT S.A. is a car manufacturer based in Catalonia. It is currently a subsidiary of the German Volkswagen group. SEAT is positioned in the market and in the Volkswagen Group as a manufacturer with a youth and sports profile.

SEAT is interested in the development of ARCO, a tool based on a Machine Learning methodology known as Proper Generalized Decomposition (PGD) to parametrically solve the NVH (Noise, Vibration and Harshness) analysis of a body characterized by material and/or geometric design variables. NVH test behavior plays a critical role in ensuring that the product meets the noise and vibration criteria, thus improving comfort, quality and customer satisfaction.

This tool is the natural continuation of previous developments made in Matlab between SEAT and LaCàN. LaCàN has the ePGD library that is specifically designed to solve problems like this. The code of the ePGD library has been developed by inLab FIB using C++ and Python, and scaling the solution (in parallel) with PETSC and OpenMPI.

SEAT has contacted inLab FIB and LaCàN to raise the possibility of developing a joint project to scale the use of ARCO and the ePGD library, currently validated for simple models, to solve NVH parametric problems with a realistic car geometry. The developed method should allow in the future a generalization of the process for other geometries and parameters.