Show simple item record

dc.contributor.advisorMulder, E.
dc.contributor.authorAdetula, Bolade Adewaleen_US
dc.date.accessioned2012-02-17T08:33:54Z
dc.date.available2012-02-17T08:33:54Z
dc.date.issued2011en_US
dc.identifier.urihttp://hdl.handle.net/10394/5561
dc.descriptionThesis (M.Sc. Engineering Sciences (Nuclear Engineering))--North-West University, Potchefstroom Campus, 2011.
dc.description.abstractCalculations of reactor parameters of interest (such as neutron multiplication factors, decay heat, reaction rates, etc.), are often based on models which are dependent on groupwise neutron cross sections. The uncertainties associated with these neutron cross sections are propagated to the final result of the calculated reactor parameters. There is a need to characterize this uncertainty and to be able to apportion the uncertainty in a calculated reactor parameter to the different sources of uncertainty in the groupwise neutron cross sections, this procedure is known as sensitivity analysis. The focus of this study is the application of a modified global sensitivity analysis technique to calculations of reactor parameters that are dependent on groupwise neutron cross–sections. Sensitivity analysis can help in identifying the important neutron cross sections for a particular model, and also helps in establishing best–estimate optimized nuclear reactor physics models with reduced uncertainties. In this study, our approach to sensitivity analysis will be similar to the variance–based global sensitivity analysis technique, which is robust, has a wide range of applicability and provides accurate sensitivity information for most models. However, this technique requires input variables to be mutually independent. A modification to this technique, that allows one to deal with input variables that are block–wise correlated and normally distributed, is presented. The implementation of the modified technique involves the calculation of multi–dimensional integrals, which can be prohibitively expensive to compute. Numerical techniques specifically suited to the evaluation of multidimensional integrals namely Monte Carlo, quasi–Monte Carlo and sparse grids methods are used, and their efficiency is compared. The modified technique is illustrated and tested on a two–group cross–section dependent problem. In all the cases considered, the results obtained with sparse grids achieved much better accuracy, while using a significantly smaller number of samples.en_US
dc.publisherNorth-West University
dc.subjectMonte Carloen_US
dc.subjectQuasi-Monte Carloen_US
dc.subjectSparse gridsen_US
dc.subjectGlobal sensitivity analysisen_US
dc.subjectGroupwiseen_US
dc.subjectNeutron cross-sectionen_US
dc.titleGlobal sensitivity analysis of reactor parametersen
dc.typeThesisen_US
dc.description.thesistypeMastersen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record