Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Physics and Astronomy

First Advisor

Mark L. Williams


This research involved developing an expert system that allows a nuclear fuel engineer to quickly provide answers to strategic nuclear fuel management questions, which are typically broad based. Current nuclear fuel analysis research concentrates on getting more accurate and precise answers at the expense of using large computer programs to get answers that are too specific to answer the broad based questions. The expert system brings together several artificial intelligence techniques to allow a nuclear fuel engineer to consider several scenarios in a general way in order to quickly answer the fuel management questions asked. The expert system is based upon a hierarchy of several abstraction levels using a constraint propagation system at the lowest level. The constraint propagation system prevents a novice nuclear fuel engineer from studying a scenario with input conditions that contradict standard nuclear fuel management relationships. The other abstraction levels include generic number representations, generic mathematical operators, and generic relationships for economic analysis. The highest level of the hierarchy is the knowledge base for nuclear fuel analysis of the equilibrium nuclear fuel cycle. The simplicity of adding other number representations to the expert system is demonstrated by implementing an interval number representation. Since the mathematical operators used at the knowledge domain level are generic, any new number representations, such as fuzzy numbers, could be added without having to change the basic domain knowledge. An example session shows how the system can be used to provide guidance to a nuclear fuel analyst in search of a good nuclear fuel management strategy. By using the interval number representation, the example session includes a simple sensitivity study on how some of the input variables' uncertainty affects the objective variable's value.