Estimation method of systematic uncertainties in Monte Carlo particle transport simulation based on analysis of variance

Hashimoto, Shintaro  ; Sato, Tatsuhiko   

Particle transport simulations based on the Monte Carlo method have been applied to shielding calculations. Estimation of not only statistical uncertainty related to the number of trials but also systematic one induced by unclear physical quantities is required to confirm the reliability of calculated results. In this study, we applied a method based on analysis of variance to shielding calculations. We proposed random- and three-condition methods. The first one determines randomly the value of the unclear quantity, while the second one uses only three values: the default value, upper and lower limits. The systematic uncertainty can be estimated adequately by the random-condition method, though it needs the large computational cost. The three-condition method can provide almost the same estimate as the random-condition method when the effect of the variation is monotonic. We found criterion to confirm convergence of the systematic uncertainty as the number of trials increases.