Improving system modeling accuracy with Monte Carlo codes
- Idaho State Univ., Pocatello, ID (United States)
The use of computer codes based on Monte Carlo methods to perform criticality calculations has become common-place. Although results frequently published in the literature report calculated k{sub eff} values to four decimal places, people who use the codes in their everyday work say that they only believe the first two decimal places of any result. The lack of confidence in the computed k{sub eff} values may be due to the tendency of the reported standard deviation to underestimate errors associated with the Monte Carlo process. The standard deviation as reported by the codes is the standard deviation of the mean of the k{sub eff} values for individual generations in the computer simulation, not the standard deviation of the computed k{sub eff} value compared with the physical system. A more subtle problem with the standard deviation of the mean as reported by the codes is that all the k{sub eff} values from the separate generations are not statistically independent since the k{sub eff} of a given generation is a function of k{sub eff} of the previous generation, which is ultimately based on the starting source. To produce a standard deviation that is more representative of the physical system, statistically independent values of k{sub eff} are needed.
- OSTI ID:
- 436941
- Report Number(s):
- CONF-9606116-; ISSN 0003-018X; TRN: 96:005275-0174
- Journal Information:
- Transactions of the American Nuclear Society, Vol. 74; Conference: Annual meeting of the American Nuclear Society (ANS), Reno, NV (United States), 16-20 Jun 1996; Other Information: PBD: 1996
- Country of Publication:
- United States
- Language:
- English
Similar Records
Apparent Monte Carlo Source Convergence Problem with BWR Fuel Depleted with Partial Control Blade Insertion
Improving variance estimation in Monte Carlo eigenvalue simulations