Parametric Analysis of Gas Diffusion Layer on PEMFC Performance Using Design of Experiment Method

Abstract

The water saturation in the gas diffusion electrodes of proton exchange membrane fuel cell (PEMFC) significantly impacts on the cell performance, but it is difficult to directly observe. The physical properties of gas diffusion electrodes, such as porosity and thermal conductivity, effects on the water saturation, but these physical properties are difficult to vary in the hardware-based experiment. Thus, the model-based engineering is significantly important to optimize the design in this area. A two-phase computational fluid dynamic (CFD) based model of PEMFC has been developed with Lattice Boltzmann Method (LBM) and validated [1, 2]. This multi-scale direct model can simulate the water saturation in the gas diffusion porous layers and predict the cell performance with co-simulation technique [3] as shown in the figure 1. The model is exercised through a rigorous design of experiment (DOE) to study the impact of the gas diffusion layers’ physical properties as well as operating conditions on the cell performance with the goal of enabling structural and operational optimization through model-based experiments. The L27 orthogonal array in the Taguchi Methods [4] was employed for the DOE. The physical properties (porosity of the gas diffusion layer (GDL) porosity, thermal conductivities of GDL and micro porous layer (MPL) and water contact angle of GDL) and operating conditions (operational temperature and current density) were allocated in the L27 array as the control factors. Since the high reproducibility is expected in the model-based experiments, the analysis of test-to-test variance was skipped. Table 1 shows a variance analysis among control factors at the current density of 1.5 A/cm2. The P-values of each control factor are very low except the water contact angle of GDL. Thus, the contact angle which represent the wettability of GDL can be insensitive for the cell performance. Figure 2 shows the results of an analysis of variance (ANOVA) study for the cell voltage as the output. Porosity and thermal conductivities of the GDL and the MPL exhibit different trend of sensitivities. It offers an opportunity to optimize them in the cell design.