Uncertainty Underprediction in Monte Carlo Eigenvalue Calculations

Mervin, B.; Mosher, Scott W.; Wagner, John C.; Maldonado, G. Ivan

doi:10.13182/nse11-104

Cited by 11 publications

(6 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This requires determination of three MCNP parameters, including number of skipped cycles (NSK), number of active cycles (NAC), and number of histories per cycle (NPS). Additionally, the variance of the fission neutron source may be significantly biased by the presence of cycle-tocycle correlation, which is difficult to detect and may cause MCNP reported variance to be underestimated [8]. This phenomenon arises from the fact that the power method [1] (used for eigenvalue Monte Carlo calculations) computes the fission source at iteration k + 1 based on the distribution at iteration k. The random numbers generated by the pseudo-random number generator (PRNG) for a certain iteration will be affected by the previous cycles, i.e., correlated.…”

Section: Determination Of Eigenvalue Parameters In Mcnpmentioning

confidence: 99%

“…In addition to the aforementioned analyses, based on a select parameters combination we performed repetition of the same run with different PRNG seeds in order to identify the presence of possible cycle-to-cycle correlation through the approach showed in [8]. From the repetition of the run, the average MCNP calculated variance is calculated as follows:…”

Section: Analysis Approach For the Establishment Of A Reference Mcnp mentioning

confidence: 99%

“…In this way, a "real" standard deviation is calculated and compared to the average MCNP uncertainty, as described in Section 4.2 and in [8]. This study was performed with smaller criticality parameters (NPS=1E6, NSK=300, and NAC=500) with respect to the study done in Section 5.1.…”

Section: Cycle-to-cycle Correlation Investigationmentioning

confidence: 99%

“…As discussed in References [8,9], cycle-to-cycle correlation is related to the method itself and it is not affected by the use of a "better" set of MCNP parameters. Therefore, we can compute the adjusted relative uncertainties for the NAC, NSK, and NPS parametric analyses (Section 5.1), multiplying each of the tally uncertainties by the weighted average ratios, f σ,wgt .…”

Section: Icrs-13 and Rpsd-2016mentioning

confidence: 99%

See 3 more Smart Citations

Evaluation of RAPID for a UNF cask benchmark problem

2017

View full text Add to dashboard Cite

Abstract. This paper examines the accuracy and performance of the RAPID (Real-time Analysis for Particle transport and In-situ Detection) code system for the simulation of a used nuclear fuel (UNF) cask. RAPID is capable of determining eigenvalue, subcritical multiplication, and pin-wise, axially-dependent fission density throughout a UNF cask. We study the source convergence based on the analysis of the different parameters used in an eigenvalue calculation in the MCNP Monte Carlo code. For this study, we consider a single assembly surrounded by absorbing plates with reflective boundary conditions. Based on the best combination of eigenvalue parameters, a reference MCNP solution for the single assembly is obtained. RAPID results are in excellent agreement with the reference MCNP solutions, while requiring significantly less computation time (i.e., minutes vs. days). A similar set of eigenvalue parameters is used to obtain a reference MCNP solution for the whole UNF cask. Because of time limitation, the MCNP results near the cask boundaries have significant uncertainties. Except for these, the RAPID results are in excellent agreement with the MCNP predictions, and its computation time is significantly lower, 35 second on 1 core versus 9.5 days on 16 cores.

show abstract

Section: Determination Of Eigenvalue Parameters In Mcnpmentioning

confidence: 99%

Section: Analysis Approach For the Establishment Of A Reference Mcnp mentioning

confidence: 99%

Section: Cycle-to-cycle Correlation Investigationmentioning

confidence: 99%

Section: Icrs-13 and Rpsd-2016mentioning

confidence: 99%

See 2 more Smart Citations

Evaluation of RAPID for a UNF cask benchmark problem

2017

View full text Add to dashboard Cite

show abstract

“…The RAPID Code System [1], developed based on the MRT methodology [2] with the Fission Matrix (FM) and the adjoint function methodologies, is capable of accurately calculating 3-D detailed fission density distribution, subcritical multiplication factor, criticality eigenvalue, and detector response for a nuclear system in real-time. RAPID achieves accurate solutions, comparable to Monte Carlo, while because of its FM method it does not suffer from the eigenvalue Monte Carlo shortcomings including particles under-sampling, source biasing and cycle-to-cycle correlation [3,4,5,6,7]. Additionally, because of its pre-calculation capability, RAPID can solve complex and large problems in real-time.…”

Section: Introductionmentioning

confidence: 99%

Experimental and Computational Validation of RAPID

Roskoff

Haghighat

Mascolino

2018

Reactor Dosimetry: 16th International Symposium

View full text Add to dashboard Cite

The RAPID (Real-time Analysis for Particle transport and In-situ Detection) code system utilizes the Multi-stage Response-function Transport (MRT) approach with the Fission Matrix (FM) method for neutronics simulation of nuclear systems. RAPID performs real-time calculations by utilizing pre-calculated databases for different enrichments, burnups, and cooling times. This paper discusses the validation of RAPID using the U.S. Naval Academy's subcritical reactor (USNA-SCR) facility. Computational validation is performed by detailed comparison with an MCNP reference calculation and experimental validation is performed using both in-core and excore neutron measurements with different 3 He proportional counters. These measurements, and associated calculations have demonstrated that RAPID achieves accurate results in real-time.

show abstract