The membrane electrode assembly (MEA) of a PEM fuel cell is a layered structure consisting of membrane, cathode and anode catalyst and gas diffusion layers. The performance of the cell can be improved by improving the cell as a whole, and by optimizing each of the layers to its requirements For example, the gas diffusion layer (GDL) plays an important role in the overall water management of the cell. The GDL is typically made of non-woven carbon paper or woven carbon cloth. For the GDL it is necessary to achieve good oxygen and water transport properties and at the same time good thermal conduction and electrical conduction properties. To improve the GDL, on the one hand, one should choose the material with the best chemical properties. On the other hand, as material connectivity and pore morphology have a major impact on the properties of porous media, improving the microstructure is equally important. Finding the best microstructure experimentally is often too costly or not possible, because any change would require changes in the production process. Computer simulations help to determine the effective material properties of a layer without the need to produce it first.
The microstructure, the morphology and the connectivity of the pores are strongly dependent on the clamping pressure applied on each cell. Thus, determining the material properties for the uncompressed media is not sufficient. This talk presents a method to determine and distinguish the effective material properties of the GDL under compression and without compression. The method consists of three steps: (i) Creation of a 3D microstructure model, (ii) Modeling the compression of the media and (iii) Simulation of the effective (two-phase) material parameters.
The 3D micro-structural models are generated by stochastic methods [1]. The simplest method uses only fiber diameter, fiber orientation and porosity as input parameters and mimics a nonwoven composed of straight fibers. More enhanced models can be achieved by adding a binder virtually or by allowing the fibers to be slightly curved.
In previous work [2], we presented a reduced compression model for porous microstructures, i.e. the reconstructed GDL. The reduced compression model was not able to capture the relation between the clamping pressure and the compression ratio. This drawback can be overcome by using a finite element (FE) approach. The FE model relies on a robust mesh generation technique in order to deal with significant voxel displacements with varying compression ratios.
Using the uncompressed and compressed 3D models, the permeability, diffusivity, electric conductivity and thermal conductivity of the layers can be predicted. This is done by solving the appropriate partial differential equations on the microstructures [3, 4]. For example, the Stokes equations are solved to obtain the effective permeability of the layer. Two-phase parameters are obtained by combining the pore morphology method and single-phase simulations [4].
The method allows to compare different microstructures and to study the effect of compression on them.