Purpose:

For some time, various research communities having been dealing with the topic of epistemic uncertainty, i.e., uncertainty due to lack of knowledge. During the last several years, the risk assessment community has begun to make a clear distinction between aleatory uncertainty, i.e., uncertainty due to a random process, and epistemic uncertainty. With this distinction, additional focus is being placed on the question of the appropriateness of using traditional probability theory to represent epistemic uncertainty in reliability engineering and risk analyses.

The Epistemic Uncertainty Project was formed at Sandia National Laboratories to investigate the applicability and usefulness of some of the modern mathematical theories for the representation of different types of uncertainty. Some of the theories of interest are interval analysis, fuzzy set theory, possibility theory, evidence (Dempster-Shafer) theory, and imprecise probability theory. Most of these theories are in an early stage of development relative to classical probability theory and Bayesian estimation. Although there are still many theoretical and applied research questions about these theories, the purpose of the Epistemic Uncertainty Project is to evaluate, develop and apply the most useful theories to reliability engineering, risk analysis, and system safety assessment.