1 - Uncertainty quantification in materials modeling

Abstract

Integrated Computational Materials Engineering (ICME) has been widely articulated and pursued over the past decade. Recognition of the need to ensure reliability and robustness of simulation predictions subject to various sources of uncertainty is central to ICME. Discrepancy between models and the physical world, as well as numerical approximation to facilitate computation, is the major source of model form uncertainty. Data fitting and model calibration processes introduce additional input uncertainty owing to systematic and random errors inherited in experimental data. Various uncertainty quantification (UQ) methods can be applied to assess the impacts of uncertainty. However, given the high cost of simulations, dimensionality is the major challenge for the application of UQ methods. In this chapter, sources of uncertainty in materials modeling and simulation are discussed. An overview is provided regarding major UQ methods and their applications to materials modeling. Their applications in simulation-assisted materials design are also reviewed.