The gas diffusion layer (GDLs) is a carbon fiber network in proton exchange membrane fuel cells (PEMFCs). This layer plays a critical role in the performance by facilitating the transport of reactant gases and providing mechanical support to the system. Understanding the properties of GDLs is essential for optimizing PEMFC efficiency and its durability. This work proposes a comprehensive characterization approach based on several descriptors beyond the obvious ones, like porosity and fiber density. Currently, numerical modelling regarding GDL is focused on either ordered systems such as woven carbon cloth or disordered ones, like non-woven carbon paper. Therefore, an attempt will be made to properly quantify the disorder within GDL, namely beyond the distribution of fibers' orientation, length, diameter and their moments. This will then be applied to explore its implications for gas flow and transport phenomena. The electrical contact resistance (ECR) between the GDL and the bipolar plate (BPP) is one of the biggest resistances in a PEMFC, which leads to losses in electrical power. The aim of the study is to develop a numerical scheme able to simulate the interface between the GDL and the BPP in order to reduce the ECR below a targeted value. Thus, parameters that quantify the roughness of contacting surfaces such as density of summits, curvature of asperities and root mean squared roughness will be also considered. In a future work, all these descriptors will then be employed with machine learning, evolutionary algorithms and coupled with lattice Boltzmann method and Greenwood-Williamson extended contact model to calculate the ECR. This will allow to gain deeper insights into the complex structure-function relationships governing GDL performance in PEMFCs.