Return to search

Neutron transport benchmarks for binary stochastic multiplying media : planar geometry, two energy groups

Benchmark calculations are performed for neutron transport in a two material
(binary) stochastic multiplying medium. Spatial, angular, and energy dependence
are included. The problem considered is based on a fuel assembly of a common
pressurized water nuclear reactor. The mean chord length through the assembly is
determined and used as the planar geometry system length. According to assumed
or calculated material distributions, this system length is populated with alternating
fuel and moderator segments of random size. Neutron flux distributions are
numerically computed using a discretized form of the Boltzmann transport equation
employing diffusion synthetic acceleration. Average quantities (group fluxes
and k-eigenvalue) and variances are calculated from an ensemble of realizations
of the mixing statistics. The effects of varying two parameters in the fuel, two
different boundary conditions, and three different sets of mixing statistics are assessed.
A probability distribution function (PDF) of the k-eigenvalue is generated
and compared with previous research. Atomic mix solutions are compared with
these benchmark ensemble average flux and k-eigenvalue solutions.
Mixing statistics with large standard deviations give the most widely varying
ensemble solutions of the flux and k-eigenvalue. The shape of the k-eigenvalue PDF
qualitatively agrees with previous work. Its overall shape is independent of variations
in fuel cross-sections for the problems considered, but its width is impacted
by these variations. Statistical distributions with smaller standard deviations alter
the shape of this PDF toward a normal distribution. The atomic mix approximation
yields large over-predictions of the ensemble average k-eigenvalue and under-predictions
of the flux. Qualitatively correct flux shapes are obtained, however.
These benchmark calculations indicate that a model which includes higher statistical
moments of the mixing statistics is needed for accurate predictions of binary
stochastic media k-eigenvalue problems. This is consistent with previous findings. / Graduation date: 2005

Identiferoai:union.ndltd.org:ORGSU/oai:ir.library.oregonstate.edu:1957/29748
Date10 March 2005
CreatorsDavis, Ian M. (Ian Mack)
ContributorsPalmer, Todd S.
Source SetsOregon State University
Languageen_US
Detected LanguageEnglish
TypeThesis/Dissertation

Page generated in 0.002 seconds