Global Monte Carlo Simulation with High Order Polynomial Expansions
The functional expansion technique (FET) was recently developed for Monte Carlo simulation. The basic idea of the FET is to expand a Monte Carlo tally in terms of a high order expansion, the coefficients of which can be estimated via the usual random walk process in a conventional Monte Carlo code. If the expansion basis is chosen carefully, the lowest order coefficient is simply the conventional histogram tally, corresponding to a flat mode. This research project studied the applicability of using the FET to estimate the fission source, from which fission sites can be sampled for the next generation. The idea is that individual fission sites contribute to expansion modes that may span the geometry being considered, possibly increasing the communication across a loosely coupled system and thereby improving convergence over the conventional fission bank approach used in most production Monte Carlo codes. The project examined a number of basis functions, including global Legendre polynomials as well as “local” piecewise polynomials such as finite element hat functions and higher order versions. The global FET showed an improvement in convergence over the conventional fission bank approach. The local FET methods showed some advantages versus global polynomials in handling geometries with discontinuous material properties. The conventional finite element hat functions had the disadvantage that the expansion coefficients could not be estimated directly but had to be obtained by solving a linear system whose matrix elements were estimated. An alternative fission matrix-based response matrix algorithm was formulated. Studies were made of two alternative applications of the FET, one based on the kernel density estimator and one based on Arnoldi’s method of minimized iterations. Preliminary results for both methods indicate improvements in fission source convergence. These developments indicate that the FET has promise for speeding up Monte Carlo fission source convergence.
- Research Organization:
- University of Michigan, Ann Arbor, Michigan
- Sponsoring Organization:
- USDOE - Office of Nuclear Energy, Science and Technology (NE)
- DOE Contract Number:
- FG07-04ID14607
- OSTI ID:
- 920974
- Report Number(s):
- FG07-04ID14607-FINAL; TRN: US0802148
- Country of Publication:
- United States
- Language:
- English
Similar Records
ADVANTG Results for a Proposed NDA System for Detecting Pin Diversion from SNF - Paper 120
Notice of Energy R&D Project