| Bibliographic Citation | |
| Full Text |  881 K |
|---|---|
| Title | Monte Carlo source convergence and the Whitesides problem |
| Creator/Author | Blomquist, R. N. |
| Publication Date | 2000 Feb 25 |
| OSTI Identifier | OSTI ID: 752868 |
| Report Number(s) | ANL/RA/CP-100164 |
| DOE Contract Number | W-31109-ENG-38 |
| Other Number(s) | TRN: AH200018%%300 |
| Resource Type | Conference |
| Resource Relation | Conference: 2000 ANS International Topical Meeting on Advances in Reactor Physics and Mathematics and Computation into the Next Millennium, Pittsburgh, PA (US), 05/07/2000--05/11/2000; Other Information: PBD: 25 Feb 2000 |
| Research Org | Argonne National Lab., IL (US) |
| Sponsoring Org | US Department of Energy (US) |
| Subject | 42 ENGINEERING; 11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS; CRITICALITY; RADIATION SOURCES; SAFETY ANALYSIS; EIGENVALUES; MONTE CARLO METHOD; CONVERGENCE; FISSION |
| Description/Abstract | The issue of fission source convergence in Monte Carlo eigenvalue calculations is of interest because of the potential consequences of erroneous criticality safety calculations. In this work, the authors compare two different techniques to improve the source convergence behavior of standard Monte Carlo calculations applied to challenging source convergence problems. The first method, super-history powering, attempts to avoid discarding important fission sites between generations by delaying stochastic sampling of the fission site bank until after several generations of multiplication. The second method, stratified sampling of the fission site bank, explicitly keeps the important sites even if conventional sampling would have eliminated them. The test problems are variants of Whitesides' Criticality of the World problem in which the fission site phase space was intentionally undersampled in order to induce marginally intolerable variability in local fission site populations. Three variants of the problem were studied, each with a different degree of coupling between fissionable pieces. Both the superhistory powering method and the stratified sampling method were shown to improve convergence behavior, although stratified sampling is more robust for the extreme case of no coupling. Neither algorithm completely eliminates the loss of the most important fissionable piece, and if coupling is absent, the lost piece cannot be recovered unless its sites from earlier generations have been retained. Finally, criteria for measuring source convergence reliability are proposed and applied to the test problems. |
| Country of Publication | United States |
| Language | English |
| Format | Medium: P; Size: 13 pages |
| Availability | INIS; OSTI as DE00752868 To purchase this media from NTIS, click here |
| System Entry Date | 2008 Feb 05 |
| Document Discussions | |
Top | |
Last Updated: 03/16/2009