Quantification of Model-Form, Predictive, and Parametric Uncertainties in Simulation-Based Design
Ramana Grandhi (Advisor), Erwin Johnson (Committee Member), Raymond Kolonay (Committee Member), Donald Kunz (Committee Member), Ravi Penmetsa (Committee Member), Joseph Slater (Committee Member)
Doctor of Philosophy (PhD)
Traditional uncertainty quantification techniques in simulation-based analysis and design focus upon on the quantification of parametric uncertainties-inherent natural variations of the input variables. This is done by developing a representation of the uncertainties in the parameters and then efficiently propagating this information through the modeling process to develop distributions or metrics regarding the output responses of interest. However, in problems with complex or newer modeling methodologies, the variabilities induced by the modeling process itself-known collectively as model-form and predictive uncertainty-can become a significant, if not greater source of uncertainty to the problem. As such, for efficient and accurate uncertainty measurements, it is necessary to consider the effects of these two additional forms of uncertainty along with the inherent parametric uncertainty. However, current methods utilized for parametric uncertainty quantification are not necessarily viable or applicable to quantify model-form or predictive uncertainties. Additionally, the quantification of these two additional forms of uncertainty can require the introduction of additional data into the problem-such as experimental data-which might not be available for particular designs and configurations, especially in the early design-stage. As such, methods must be developed for the efficient quantification of uncertainties from all sources, as well as from all permutations of sources to handle problems where a full array of input data is unavailable. This work develops and applies methods for the quantification of these uncertainties with specific application to the simulation-based analysis of aeroelastic structures.
Department or Program
Ph.D. in Engineering
Year Degree Awarded
Copyright 2011, all rights reserved. This open access ETD is published by Wright State University and OhioLINK.