Abstract

This report introduces some key concepts in uncertainty quantification for engineering applications. It reviews some of the common approaches to dealing with uncertainties in computational models and some of the formal, and less formal, approaches for constructing models of uncertainty with a clear statement of the assumptions and caveats that are explicitly or implicitly being made in their application. The report argues for a clearer appreciation of the assumptions that are made in any analysis, and offers ways to relax assumptions that may limit analyses for computational simplicity or unjustified convention. The report begins by introducing the philosophical difference between types of uncertainty with a focus on its practical implications, and then builds upon these definitions to distinguish between methodological approaches to dealing with this uncertainty. Finally, the report outlines the essential uncertainty-aware questions any engineer must ask themselves before they develop uncertainty quantification analyses, and touches on some of the recent and more classical approaches from the literature to answer these essential questions. This report is intended to be a primer on models of uncertainty, and the methods for exploring their effect. We survey the most important aspects of uncertainty modelling and point the reader to references to further investigate these key concepts. Portions of grey text interspersed throughout this report provide more detail about the section’s topic. This report is the first in a series. The next report will focus on the development of advanced solution sequences and the process of implementing uncertainty quantification workflows in real world applications. Subsequent reports will focus on key topics in uncertainty quantification: algorithms for uncertainty propagation, verification and validation, and calibration. 

This is v. 1.1.
Changes from previous version:

• fixed a typo in the fourth paragraph of "Further discussion - Controversies in uncertainty analysis"

• fixed a rendering issue in the methods table on page 50

DAWS Synopsis on Uncertainty Quantification

The DAWS Synopsis on UQ presents the essence of the foundations, justification and methods of UQ. It is a meant to serve as a starter for people making their first steps in the field. More detailed information on all of the topics discussed in the synopsis can be found in the DAWS Uncertainty Framework above.