Validation of an adaptive order Finite Element model for transmission loss simulation of trimmed automotive structures

Vehicle electrification and increasing demands for driving comfort present significant challenges for designing effective noise control treatments in modern vehicles. Lightweight, low-emission designs often compromise acoustic efficiency. One way of compensating this is through the use of multi-layer ‘trim’ material configurations to mitigate noise across a wider frequency range. Traditional 3D finite element models, while accurate and even needed to capture the full dynamic behaviour, become computationally prohibitive for complex automotive structures like firewalls, which feature intricate shapes, high curvature, and material compression. This computational burden limits design exploration and timely noise performance predictions. This paper introduces the use of an innovative adaptive higher-order finite element method to address these challenges. The method utilizes a fixed finite element discretization of trim components and surrounding acoustic fluid, incorporating an automated refinement strategy. This strategy locally enhances accuracy based on the frequency of interest and material properties, ensuring an optimal balance between calculation time and prediction accuracy at each analysis frequency. We detail the application of this framework to evaluate the sound transmission loss (STL) of automotive firewalls, including the effect of poro-elastic and viscoelastic soundproofing materials. In collaboration with Mercedes-Benz AG, an adaptive order finite element model was developed for a STL test setup. We present simulation results for both untrimmed and trimmed configurations, comparing them against experimental data for the panel STL levels and relative improvements due to a sound package modification. The findings demonstrate the method's accuracy, efficiency, and applicability to real-world automotive engineering problems, also shed light on the trade-offs between model idealization and fidelity of the digital twin.