Hydrogen fuel cells have held out the promise of clean, sustainable power generation for decades, but have failed to deliver on that potential. Inefficiencies in research and development work can be overcome to increase the rate of new knowledge acquisition in this field. A number of medical and engineering disciplines utilise a wide variety of statistical tools in their research to achieve this same end, but there has been little adoption of such statistical approaches within the fuel cell research community. This research undertakes a design of experiments (DoE) approach to the analysis of multiply-covarying (M-ANOVAR) factors by using historic data, and direct experimental work, on a wide variety of polymer electrolyte membrane fuel cells (PEMFCs) cathode gas diffusion media (GDM) and dual layered catalyst structures. This research developed a gradient of polarisation regions' approach; a method for making robust numerical comparisons between large numbers of samples based on polarisation curves, while still measuring the more usual peak power of the PEMFC. The assessment of polarisation gradients was completed in a statistically robust fashion that enabled the creation of regression models of GDMs for multiple input and multiple output data sets. Having established the multivariate method; a set of possibly co-varying factors, a DoE approach was used to assess GDM selection, dual layered catalyst structures and degradation of membrane electrode assembly (MEA) performance over time. Degradation studies monopolise resources to be monopolised for protracted periods. M-ANOVAR allows the addition of other factors in the study, and the total efficiency of the degradation experiment is increased. A 20% reduction in the number of samples to be tested was achieved in the case study presented in this thesis (compared to the usual one factor at a time (OFAT) approach). This research highlights the flexibility and efficiency of DoE approaches to PEMFC degradation experimentation. This research is unique in that it creates catalyst ink formulations where the variation in catalyst loading in each sub-layer of the catalyst layer (CL) was achieved by having a different concentration of the catalyst material on the carbon supports. The final M-ANOVAR analysis indicates a simple average of the individual responses was appropriate for the experiments undertaken. It was shown that low concentration dual layer catalysts on paper GDMs have improved performance compared to paper GDMs with uniform, single layer catalysts: Demonstrating reduced platinum concentrations to achieve equivalent open cell performance. The time to peak power during testing (how long after starting the test it takes to achieve the maximum performance in the cell) was strongly impacted by GDM selection. Furthermore, there was a strong suggestion that previously published results crediting a change in performance due to a single layer, or multi-layered catalyst structures may, in fact, have been due to the selection of GDM used in the experiment instead.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:747914 |
| Date | January 2017 |
| Creators | McCarthy, Nicholas |
| Publisher | Loughborough University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://dspace.lboro.ac.uk/2134/26273 |
Page generated in 0.0017 seconds