| Abstract Scope |
Calibration is a critical step in the development of predictive material constituent physics models, yet the appropriate strategy depends strongly on specific details of models, available experimental data, and intended application. This presentation begins with a broad overview of calibration approaches employed across a diverse portfolio of computational physics research efforts, highlighting common principles as well as important differences in methodology across application domains. The discussion then focuses on recent advances in the calibration of crystal plasticity and
macroscopic high-rate strength constitutive models. These examples employ Bayesian inference together with automated computational workflows to efficiently integrate simulations, experimental data, parameter estimation, and uncertainty analysis. Beyond improving the efficiency of calibration, these approaches provide quantitative insight into parameter
identifiability, model sensitivity, and the adequacy of underlying physical assumptions. Rather than viewing calibration solely as a means of estimating model parameters, these
examples demonstrate how Bayesian inference and automated workflows can be used to identify model deficiencies, quantify uncertainty, and guide the development of improved constitutive
models. Calibration thus becomes an integral part of the scientific process, providing a systematic framework for learning from experiments and advancing predictive materials modeling. |