Kyle J. Lange

Pore Scale Simulation of Transport and Electrochemical Reactions in Reconstructed PEMFC Catalyst Layers

A mesoscale simulation is developed to simulate transport and electrochemistry in a small section of a proton exchange membrane fuel cell (PEMFC) cathode catalyst layer. Oxygen, proton, and electron transport are considered in the model. Many simulations are run with a wide variety of different parameters on stochastically reconstructed microstructures with a resolution of 2 nm. Knudsen diffusion plays an important role in limiting the transport of oxygen through the catalyst layer. Using larger carbon spheres in the catalyst layer increases the effective diffusivity of oxygen through the catalyst layer. The effective proton conductivity increases when larger spheres are used, a normal distribution of spheres is used, or a higher overlap tolerance is used. Increasing the overlap tolerance or overlap probability results in an increase in the effective electron conductivity. When electrochemical reactions are considered in a part of the catalyst layer that is close to the gas diffusion layer, the critical parameter that determines oxygen consumption is the carbon sphere radius. Oxygen consumption at a given carbon volume fraction is larger in microstructures containing spheres with smaller radii, because there is more surface area available for electrochemical reactions.

Using sensitivity derivatives for design and parameter estimation in an atmospheric plasma discharge simulation

The problem of applying sensitivity analysis to a one-dimensional atmospheric radio frequency plasma discharge simulation is considered. A fluid simulation is used to model an atmospheric pressure radio frequency helium discharge with a small nitrogen impurity. Sensitivity derivatives are computed for the peak electron density with respect to physical inputs to the simulation. These derivatives are verified using several different methods to compute sensitivity derivatives. It is then demonstrated how sensitivity derivatives can be used within a design cycle to change these physical inputs so as to increase the peak electron density. It is also shown how sensitivity analysis can be used in conjunction with experimental data to obtain better estimates for rate and transport parameters. Finally, it is described how sensitivity analysis could be used to compute an upper bound on the uncertainty for results from a simulation.

Sensitivity Derivatives for Plasma Discharge Simulations

The problem of applying sensitivity analysis to plasma discharge simulations is considered. A fluid model of a low-pressure radio frequency helium discharge is used to demonstrate how sensitivity derivatives can be computed for an unsteady periodic solution. The derivations of forward mode direct differentiation and the reverse mode adjoint method are presented. Good agreement is shown between sensitivity derivatives computed by these methods and those computed by finite differences. It is then demonstrated how sensitivity derivatives can be used as part of a design cycle to increase or decrease a given cost function. Finally, it is demonstrated how sensitivity derivatives can be used to compute error bounds for a given cost function.