(Invited) Model Guided Design and Analysis of Experiment

Abstract

Batteries are complex, with phenomena emanating from multiple length scales that may impact performance and life. Advances thus require multiscale experimental inquiries, and mathematical models, including multiscale models, may be employed to design, analyze and integrate studies. In early-stage research efforts, close collaboration with experimental efforts may result both in dramatically improved model fidelity and in more optimal utilization of experimental resources. We present approaches to augment physics-based models of Li-ion cathodes with applied statistics algorithms based on Markov-chain Monte Carlo simulations. Several examples are illustrated. The use of model regularization algorithms as a tool to efficiently introduce new physics in cathode-performance and degradation models is demonstrated. Additionally, model selection algorithms are demonstrated within the context of studies of mesoscale transport processes. Identification of the better model leads to an understanding of the dominant length scale in terms of impedance. Such tools can inform decisions in terms of where to focus experimental resources. Finally, the model-guided design of experiment (MGDOE) is discussed. Specifically, the use of simulation tools to identify experimental approaches to minimize parameter uncertainty is discussed, and the use of MGDOE for synchrotron-based operando experiments is considered.