Stratified source-sampling techniques for Monte Carlo eigenvalue analysis.
In 1995, at a conference on criticality safety, a special session was devoted to the Monte Carlo ''Eigenvalue of the World'' problem. Argonne presented a paper, at that session, in which the anomalies originally observed in that problem were reproduced in a much simplified model-problem configuration, and removed by a version of stratified source-sampling. In this paper, stratified source-sampling techniques are generalized and applied to three different Eigenvalue of the World configurations which take into account real-world statistical noise sources not included in the model problem, but which differ in the amount of neutronic coupling among the constituents of each configuration. It is concluded that, in Monte Carlo eigenvalue analysis of loosely-coupled arrays, the use of stratified source-sampling reduces the probability of encountering an anomalous result over that if conventional source-sampling methods are used. However, this gain in reliability is substantially less than that observed in the model-problem results.
- Research Organization:
- Argonne National Lab., IL (US)
- Sponsoring Organization:
- US Department of Energy (US)
- DOE Contract Number:
- W-31109-ENG-38
- OSTI ID:
- 10648
- Report Number(s):
- ANL/RA/CP-95657; TRN: US0103790
- Resource Relation:
- Conference: International Conference on the Physics of Nuclear Science and Technology, Long Island, NY (US), 10/05/1998--10/08/1998; Other Information: PBD: 10 Jul 1998
- Country of Publication:
- United States
- Language:
- English
Similar Records
Monte Carlo Source Convergence and the Whitesides Problem
Estimation of distributions via multilevel Monte Carlo with stratified sampling