This presentation was made at CAASE18, The Conference on Advancing Analysis & Simulation in Engineering. CAASE18 brought together the leading visionaries, developers, and practitioners of CAE-related technologies in an open forum, to share experiences, discuss relevant trends, discover common themes, and explore future issues.
Recent progress in micro-scale three-dimensional (3D) characterization techniques, such as Focused Ion Beam - Scanning Electron Microscopy (FIB-SEM) and X-ray nanotomography, has brought new ways to relate material microstructure to performance/properties. Such techniques have become increasingly common for studying solid oxide fuel cell (SOFC) microstructures. Most reported reconstructions have been based on in-house cells made in controlled laboratory settings, which have been argued to have uniform, or homogeneous, microstructures. Electrochemistry models based on these average or uniform microstructures have produced good agreement for the average performance of cells from which the reconstructions were made. Nevertheless, unacceptably high degradation rates inhibit widespread commercialization of SOFCs. It is possible that degradation is more closely related to the specific internal distribution of the 3D microstructural features, rather than the mean values from a given volume. Furthermore, commercial cells, or cells manufactured in large-scale assembly lines, have shown substantially less microstructural uniformity than that reported for in-house cells. These microstructures exhibit local phase heterogeneities across many different length scales, from microns to tens of microns and more, and they are correlated to tails or outliers of distributions in transport/electrochemical properties. Thus, it is critical to study such commercial microstructures with dimensions large enough so that full distributions of performance metrics can be generated.
This work aims to quantify local distributions of transport/electrochemical properties in commercial SOFC microstructures over large length scales that can populate a significant number of outliers. An emphasis is placed on the workflow to scan, mesh, and model complex SOFC microstructures over tens of microns and beyond using the Finite Element (FE) method. Specifically, we use high-resolution (~50 nm point spacing) Xe-plasma FIB-SEM to capture microstructures with dimensions on the order of 100-200 µm. Since local electrochemistry takes place at morphological features that vary throughout microstructure, it is important to retain the morphologies when converting microstructure image data to computational domains (for computing electrochemistry). To this end, we use ScanIP and the FE add-on module in the Simpleware software platform (Synopsys, Inc., Mountain View, CA) to convert the 3D scan image data into microstructural, multi-domain and simulation-ready volumetric FE meshes that preserve surface morphologies in three-phase SOFC electrodes. Finally, we use MOOSE (Idaho National Laboratory) – an open-source FE framework designed for high-performance platforms – to simulate microstructure-based electrochemistry on a supercomputer (Joule, National Energy Technology Laboratory). Avoiding singularities at triple lines motivated converting the latter to small volumetric features (50-100 nm) so that reactions between the phases – as well as transport along the lines – can be defined. Preliminary results regarding distributions of local electrochemical parameters throughout large-scale microstructures will be presented using data analytics and statistical sampling techniques.
|Date||5th June 2018|
|Organisation||Carnegie Mellon University|