| dc.description.abstract | A proton exchange membrane fuel cell (PEMFC) is an energy-conversion device that can be used in a wide variety of applications and is considered a promising alternative to gas and diesel engines for vehicular purposes. During operation, a significant source of energy losses stem from the catalyst layer (CL) of the PEMFC, where resistance to transport and reaction mechanisms under high current density can severely reduce the power output of the device. Computational reconstructions of the CL microstructure are therefore very important in order to study the pore-scale phenomena that take place there. Conventional reconstruction techniques typically use either experimental images, which is expensive and time-consuming or use stochastic reconstruction techniques such as simulated annealing, which strive to generate a statistically accurate representation of the structure, and therefore depend on the selection of appropriate statistical correlations to optimize. Most studies to date have exclusively optimized the two-point correlation in simulated annealing despite other studies indicating it does not adequately characterize a unique microstructure by itself.
In this thesis, a hybrid stochastic reconstruction technique is presented, combining experimental images with simulated annealing. Two additional statistical correlations are considered for optimization, namely the lineal path distribution and a 2D analog of the pore-size distribution. The reconstruction technique reconstructs a CL from 2D scanning electron microscope experimental images, combining a sphere-packing approach with sphere-based simulated annealing and simultaneously optimizes the microstructure using the porosity, the two-point correlation, the lineal path distributions and the pore size distributions. Each reconstruction is analyzed by comparing its porosity, surface area, mean chord lengths and pore size distributions with those of the reference image. The study found that the most accurate reconstructions incorporate both the lineal path distribution and the pore size distributions. Pore-scale transport simulations are conducted on the reconstructions to obtain the effective thermal, ionic, and electrical conductivities in addition to the effective diffusivity, in which Knudsen effects are accounted for by using a random walk method. Estimates for the effective ionic and thermal conductivities are in good agreement with those obtained from reference image. However, effective diffusivity and effective electrical conductivity are greatly over and underestimated, respectively. This is attributed to the high sphericity of the sphere-based reconstructions. Narrow connections between sphere contribute to the low electrical conductivity, and the pore phase lacks the long, dead-end pores present in the reference image, which causes the mass diffusivity to be overestimated. | en |
