BEPU and Evaluation of Predictive Capability of Physics Simulations of CANDU Transients

Abstract of the technical paper/presentation presented at:
ANS Best-Estimate Plus Uncertainty International Conference (BEPU 2018)
May 13–19, 2018

Prepared by:
Dumitru Serghiuta,^{Footnote 1} John Tholammakkil,1 Hany Abdel-Khalik^{Footnote 2}

Abstract

The current CNSC REGDOC-2.4.1, Deterministic Safety Analysis, allows for the use of more realistic methodologies, such as the best estimate reactor analysis simulations with consideration of uncertainties. Quantification and understanding of uncertainty sources is an essential requirement of best-estimate plus uncertainties (BEPU) analysis, as it provides a reliable metric by which the quality of the predictions can be assessed. Although direct comparison against measurements provides the ultimate evidence that simulation predictions are reliable, the true value of any BEPU simulation lies in its ability to provide predictions for conditions for which measurements are unavailable. Therefore, there is a clear need to characterize (i.e., to propagate and prioritize) all sources of uncertainties, in order to reliably use the results of BEPU calculations in the various aspects of reactor design, operation and safety.

CNSC staff have initiated a study to investigate the feasibility of developing a first-of-its-kind integrated framework for uncertainty characterization (UCF), with primary application to CANDU neutronics calculations. This study has been undertaken with the aim of enhancing the CNSC's capability to independently verify safety cases that use more realistic methodologies, in particular those that rely on complex analytical simulations using 3D-neutronics thermal-hydraulic coupled computational procedures.

This paper describes the approach for implementing a reduced UFC that is focused on developing an integrated automated capability for uncertainty analysis for the reactor core simulator NESTLE-C in both steady state and transient CANDU core calculations.

The first step is to develop a complete library of diffusion theory-based group cross-sections and associated uncertainty distribution function and covariance matrices, as well as an integrated platform that includes execution scripts, sampler software, and processing software to estimate the joint probability distribution for all output results. The output results are also customized to include typical statistical quantities such as means, variances, and covariance information for user-selected core attributes. The main computer codes employed in the generation of the diffusion theory cross-sections library are SERPENT, TRITON, T2N and CRANE.

To obtain a copy of the abstract's document, please contact us at cnsc.info.ccsn@cnsc-ccsn.gc.ca or call 613-995-5894 or 1-800-668-5284 (in Canada). When contacting us, please provide the title and date of the abstract.