NNSA PRISM Seminar: "Uncertainty Quantification and Model Validation in Mechanical System...."
June 14 @ 10:00 AM - 11:00 AM - Burton Morgan Center for Entrepreneurship, Room 129
Abstract: This talk will present recent methods to address several challenges in model-based performance prediction of engineering systems. Model-based simulation becomes attractive for systems that are too large and complex for full-scale testing. However, model-based simulation involves many approximations and assumptions, and thus confidence in the simulation result is an important issue. Sources of uncertainty are both aleatory and epistemic, stemming from natural variability, information uncertainty, and modeling errors. Recent research on methods for model validation, model calibration, and prediction uncertainty quantification will be discussed. The presentation will also include methods for quantifying various types of errors and uncertainties, including model form uncertainty; methods for assessing confidence in extrapolation to usage conditions different from laboratory conditions; and methods for test design/planning to meet validation and prediction uncertainty requirements. These techniques will be illustrated with a variety of mechanics problems, including multi-level, multi-disciplinary systems, and materials damage modeling.
- Mari-Ellyn Brock