The Role of Uncertainty Quantification in Model Verification and Validation

Thursday, July 30, 2015 - 1:30pm - 5:00pm
DLR Room 143 A&B
David Riha (Southwest Research Institute)
Uncertainty quantification plays a critical role in establishing model prediction capability and credibility. Model verification and validation (V&V) frameworks have been developed for the fluid mechanics, solid mechanics, and integrated computational materials engineering (ICME) disciplines as well as organization specific guidance for NASA and the U.S. Department of Defense to address the need to confirm that model predictions are sufficiently accurate to support programmatic decisions. This tutorial will introduce concepts of model V&V and how uncertainty analysis is used to support the process. A realistic case study based on a model developed to predict the temperature dependent yield strength of nickel-based superalloys will be used to introduce concepts and demonstrate analysis strategies for identifying, quantifying, and reducing uncertainty in the model predictions. The attendees will use the NESSUS® probabilistic analysis software to model and propagate uncertainties through the yield strength model, compute global sensitivities to identify important uncertainties, and calibrate model inputs to reduce uncertainties. Practical aspects of uncertainty quantification and the communication of results for decision making will be emphasized.