PhD Defense - Valentin Miqueau

Raliability of the SEA vibro-acoustic prediction method in the context of subjective assessment of automotive acoustic comfort

Abstract :

As the interiors of new vehicles become quieter and quieter, sound quality and interior acoustic comfort have become key arguments for car manufacturers to differentiate themselves from their competitors.
At present, research into sound perception in cars relies on an experimental approach based on signals recorded on expensive prototypes.
The aim of this thesis is to study the possibility of using the numerical predictions of a model based on the SEA (Statistical Energy Analysis) method and developed by Saint-Gobain Research Compiègne, in order to generate the signals required for subjective experiments.
This model can predict the impact of glazing changes on the noise level of a vehicle exposed to environmental noise. However, its use to assess vehicle comfort has raised a number of questions:
(i) How can audio signals be generated from predicted sound pressure levels?
(ii) Can the obtained signals be used to carry out studies of acoustic comfort in the passenger compartment with the same fidelity as recordings measured on a vehicle?
These questions will be addressed by (1) setting up a method for Auralizing the results of the SEA digital model and conducting subjective experiments to compare the annoyance experienced as a function of the origin of the sounds (measured or simulated). A discrepancy in the assessments for tempered glass was then identified. A dissipation contribution from the joints, not taken into account in the model, is suspected as the cause. (2) The vibro-acoustic behaviour of tempered glass mounted in automotive joints and its sensitivity to the dissipation they contribute is therefore shown to be non-negligible. (3) An FEM model is therefore set up to predict the transmission losses characterising the complete system (glass and joints) in the SEA model.