Project Details
Projekt Print View

Surface contributions with respect to acoustic energy quantities applied to cavities

Subject Area Acoustics
Term from 2019 to 2024
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 418936727
 
Final Report Year 2024

Final Report Abstract

In this research project, a numerical method has been developed to identify sound radiating sources on vibrating surfaces in interior acoustics. The impact of these structures to a prescribed objective function are known as surface contributions. Existing methods focus on the sound pressure as the objective function. However, this entity strongly depends on the evaluated position and results, thus, in deteriorated performances in regions with low sound pressure values. Therefore, surface contributions based on the acoustic energy are implemented in this project. In a subsequent step, the surface contributions have been extended to consider the entire volume as the objective function. Lastly, an additional method has been implemented for accelerated frequency sweep analyses, which are omnipresent in studies of vibroacoustic systems. The proposed method has been implemented in a BEM framework. For this task, the BEM is preferred to the FEM, as the BEM solely requires a discretization of the enveloping surface. The acoustic energy density, comprising the (sound pressure-proportional) potential and the (particle velocity-proportional) kinetic energy density, is used as the objective function. The findings show that energy-based surface contributions yield robust predictions, particularly, in regions and frequencies with low sound pressure values. In analogy to the Gaussian integration, the acoustic energy densities at multiple positions are regarded as integration points. The volume-related surface contributions are then evaluated by computing the sum over all integration points allowing an efficient approximation of the surface contributions regarding the entire volume. Finally, a method has been developed to accelerate frequency sweep analyses of acoustic systems. For this purpose, a multi-fidelity model has been deployed combining the accuracy of a high-fidelity model with low-fidelity solutions, which are fast to obtain. This multi-fidelity model is realized by Gaussian processes, which are based on the Bayesian probability principle. By this means, uncertainties due to simplifying modeling assumptions or limited knowledge on model parameters are inherently quantified. All methods have been verified with a real-world, industrial application: the low-frequency booming noise in passenger cabins. Taken together, a robust and efficient diagnosis tool has been developed to identify surface contributions in interior acoustics. By using the acoustic energy quantities as the objective function, deeper insight into the acoustic system is gained. On top of that, a probabilistic method has been proposed for accelerated frequency sweep analyses enabling robust predictions under partial information. This paves the way for fast decision making processes, particularly, in early phases of product development cycles.

Publications

 
 

Additional Information

Textvergrößerung und Kontrastanpassung