Reconstructed fibrous media such as glass wools can be obtained as a series of elementary volumes of interest, which are filled with solid fibers. This results in a stochastic microstructural model, parameterized by the probability laws inferred from the database. Here, the fiber radii of the solid fibers are described by two different models (i) the diameter is randomly sampled for each fiber, according to a Gamma law estimated from the data; (ii) the diameter of each fiber is weighted by its volume and the empirical distribution of the weighted diameters thus defined is sampled. Quasi-static transport parameters, such as the viscous permeability, are derived from the second model involving volume weighting; whereas high-frequency transport parameters, like the viscous characteristic length, are computed from the first model. The need of two different models to characterize the microstructure and quantify the corresponding transport parameters provides insight for analysing transport processes through glass wool panels. In this talk, the implications emerging from these results will be discussed along with possible future works.