Multiscale prediction of acoustic properties for glass wools: Computational study and experimental validation.

Abstract

This work is concerned with the multiscale prediction of the transport and sound absorption properties associated with industrial glass wool samples. In the first step, an experimental characterization is performed on various products using optical granulometry and porosity measurements. A morphological analysis, based on scanning electron imaging, is further conducted to identify the probability density functions associated with the fiber angular orientation. The key morphological characterization parameters of the microstructure, which serve as input parameters of the model, include the porosity, the weighted volume diameter accounting for both lengths and diameters of the analyzed fibers (and therefore the specific surface area of the random fibrous material), and the preferred out-of-plane fiber orientation generated by the manufacturing process. A computational framework is subsequently proposed and allows for the reconstruction of an equivalent fibrous network. A fully stochastic microstructural model, parameterized by the probability laws inferred from the database, is also proposed herein. Multiscale simulations are carried out to estimate transport properties and sound absorption. With no adjustable parameter, the results accounting for ten different samples obtained with various processing parameters are finally compared with the experimental data and used to assess the relevance of the reconstruction procedures and the multiscale computations.