An Analysis on the Correlates of Nuclear Proliferation and Nuclear Energy

Subbaiah, Meyappan

An Analysis on the Correlates of Nuclear Proliferation and Nuclear Energy

dc.contributor.advisor	Chirayath, Sunil S.
dc.contributor.committeeMember	Nelson, Paul
dc.contributor.committeeMember	Fuhrmann, Matthew
dc.creator	Subbaiah, Meyappan
dc.date.accessioned	2016-09-22T19:45:33Z
dc.date.available	2018-08-01T05:57:30Z
dc.date.created	2016-08
dc.date.issued	2016-07-14
dc.date.submitted	August 2016
dc.date.updated	2016-09-22T19:45:33Z
dc.description.abstract	The study of various indicators of nuclear proliferation actions by states can identify the associated level of risk. This study expands upon previous proliferation risk work by investigating the number of Enrichment and Reprocessing facilities a state has based on various historical indicators. These indicators include: (a) Gross Domestic Product (GDP) Per Capita, (b) Nuclear Electricity Production, (c) Possession of Nuclear Weapons, (d) Superpower Alliance, (e) Technical Capabilities, (f) number of Rival ENR facilities, and (g) number of ENR facilities held by a trading partner. ENR facilities are a vital part of the nuclear fuel cycle, regardless of intent be it civilian electricity production or weapons production. The number of ENR facilities is important to measure, as this provides information regarding a state's urgency and reasoning for a weapons program. Data, from A Spatial Model of Nuclear Technology Diffusion by M. Fuhrmann and B. Tkach, is utilized to develop a predictive model. This dataset includes state data from 1945-2010, for 56 countries that had at least one operational research reactor. From the aforementioned indicators, both the number of Rival ENR facilities and number of ENR facilities held by a trading partner accounted for spatial clustering of nuclear weapons programs. Spatial clustering provided the opportunity to capture the dynamic nature of proliferation. Bayesian networks were used as the investigative tool for this study. These networks are directed acyclic graphs that provide the ability to represent conditional dependence relationships between sets of random variables. This provides the ability to use information about the state of a random variable to infer additional information about the other random variable. Bayesian networks allow for a more visual approach to developing joint distributions of all important variables that model a system. In most cases, there is a plethora of data for Bayesian networks to be constructed from. It is possible to inform these networks through expert judgement. However, due to the limited data available for nuclear weapons history, expert judgments are also required to ensure model specification. From this study, it was evident that Bayesian networks were an appropriate tool to capture the dynamics of a potential proliferation threat and the level of proliferation risk. However, due to the complexity behind nuclear weapons programs there is always an opportunity for future work. The results from this study compared favorably to the historical data from Fuhrmann and Tkach, with some potential for better prediction accuracy. Refined models, with a higher validation rate with respect to historical data, can be used as a policy tool. These refined models will have the capability to forecast.	en
dc.format.mimetype	application/pdf
dc.identifier.uri	https://hdl.handle.net/1969.1/158034
dc.language.iso	en
dc.subject	nuclear security	en
dc.subject	nuclear proliferation	en
dc.title	An Analysis on the Correlates of Nuclear Proliferation and Nuclear Energy	en
dc.type	Thesis	en
dc.type.material	text	en
local.embargo.terms	2018-08-01
local.etdauthor.orcid	0000-0001-8146-4701
thesis.degree.department	Nuclear Engineering	en
thesis.degree.discipline	Nuclear Engineering	en
thesis.degree.grantor	Texas A & M University	en
thesis.degree.level	Masters	en
thesis.degree.name	Master of Science	en