Title:
Investigation and improvement of criticality calculations in MCNP5 involving Shannon entropy convergence
Investigation and improvement of criticality calculations in MCNP5 involving Shannon entropy convergence
Author(s)
Koch, David
Advisor(s)
Petrovic, Bojan
Editor(s)
Collections
Supplementary to
Permanent Link
Abstract
Criticality calculations are often performed in MCNP5 using the Shannon entropy as an indicator of source convergence for the given neutron transport problem. The Shannon entropy is a concept that comes from information theory. The Shannon entropy is calculated for each batch in MCNP5, and it has been shown that the Shannon entropy tends to converge to a single value as the source distribution converges. MCNP5 has its own criteria for when the Shannon entropy has converged and recommends a number for how many batches should be skipped; however, this value for how many batches should be skipped is often not very accurate and has room for improvement.
This work will investigate an approach for using the Shannon entropy source distribution convergence information obtained in a shorter simulation to predict the required number of generations skipped in the reference case with desired statistical precision. In several test cases, it has been found that running a lesser number of particles per batch produces a similar Shannon entropy graph when compared to running more particles per batch. Then, by appropriate adjustment through a synthetic model, one is able to determine when the Shannon entropy will converge by running fewer particles, finding the point where it converges and then using this value to determine how many batches one should skip for a given problem. This reduces computational time and any "guessing" involved when deciding how many batches to skip. Thus, the purpose of this research is to develop a model showing how one can use this concept and produce a streamlined approach for applying this concept to a criticality problem.
Sponsor
Date Issued
2014-12-05
Extent
Resource Type
Text
Resource Subtype
Thesis