This paper presents the 3D carbon cloth gas diffusion layer (GDL) to predict transport behaviors of anisotropic structure properties. A statistical characterization and stochastic reconstruction method is established to construct the 3D micro-structure using the data from the true materials. Statistics of the many microstructure characteristics, such as porosity, pore size distribution, and shape of the void, are all quantified by image-based characterization. Furthermore, the stochastic reconstruction algorithm is proposed to generate random and anisotropic 3D microstructure models. The proposed method is demonstrated by some classical simulation prediction and to give the evaluation of the transport properties. Various reconstructed GDLs are also generated to demonstrate the capability of the proposed method. In the end, the adapted structure properties are offered to optimize the carbon cloth GDLs.
Unsupervised image segementation by stochastic reconstruction
