Speaker
Description
Nuclear analyses performed using Monte Carlo (MC) radiation transport codes are affected by stochastic uncertainty arising from the probabilistic sampling of particle histories. While formal mathematical convergence cannot be guaranteed, MC transport codes provide statistical estimators—such as the relative stochastic uncertainty—that allow the assessment of result reliability. Compliance with the tally statistical tests, including sufficiently low stochastic uncertainties, is therefore a mandatory requirement to ensure the numerical robustness of the simulation results.
Complex nuclear analyses may require the coupling of two Monte Carlo (MC) radiation transport simulations. The most widely adopted two–MC-step methodology is the Rigorous Two-Step (R2S) approach, commonly used for shutdown dose rate calculations. This methodology sequentially couples an MC neutron transport calculation, an activation calculation, and an MC photon transport calculation. Consequently, the statistical uncertainty introduced in the initial neutron transport stage propagates through the subsequent steps, affecting the photon transport results. However, the standard R2S methodology does not provide quantitative information on the impact of neutron stochastic uncertainty on the final response, thereby missing relevant information regarding the numerical reliability of the calculated results. This limitation is usually addressed by increasing the number of particles simulated during the neutron transport calculation in order to reduce its statistical uncertainty as much as possible, at the expense of a significantly increased computational cost.
Existing methodologies developed to address this issue require calculating the neutron flux covariance matrix to propagate the stochastic uncertainty in the neutron flux to the response. However, the size of this matrix makes its explicit calculation impractical for realistic applications. At the same time, simplified assumptions for its structure can lead to a significant overestimation of the resulting uncertainty, thereby increasing the overall computational cost.
In this work, we present the R2SUNED implicit stochastic uncertainty propagation scheme, which overcomes these limitations. The proposed scheme enables an efficient and accurate assessment of the contribution of neutron stochastic uncertainty to the final response, without relying on prohibitive covariance matrix calculations. The methodology has been applied to a JET shutdown dose rate analysis, demonstrating its applicability to real fusion-relevant problems and its contribution to improving the numerical reliability and robustness of R2S-based results.
| Formatted abstract uploaded? | Done. |
|---|