Integrating best estimate plus uncertainty analysis into model-based systems engineering

Abstract

The licensing of new nuclear reactors after the Fukushima accident presents significant challenges due to the complexity of nuclear systems and the stringent regulatory requirements involved. Model-based systems engineering (MBSE) has emerged as a useful approach for managing the development of such complex systems, while Best Estimate Plus Uncertainty (BEPU) methodologies have proven valuable within regulatory frameworks for safety evaluation. However, digital models and databases that are needed to provide evidence that the system meets the specified requirements are usually isolated in discipline-specific data repositories. To address this challenge, this article proposes a model breakdown structure (MBS) methodology, using a set of interconnectable models to seamlessly integrate MBSE, computer-aided engineering (CAE) models, and BEPU analysis. The Brazilian Multipurpose Reactor (RMB) served as the system of interest to exemplify the effectiveness of the proposed methodology. A requirement specification was linked to a finite element analysis (FEA) that estimates the peak cladding temperature in a slow loss of flow accident scenario. Additionally, key design factors are identified using design of experiments (DOE) and analysis of variance (ANOVA). Lastly, Wilks' theorem and Monte Carlo simulations are applied for uncertainty quantification. The results indicate that the 95/95 upper tolerance limit of the peak cladding temperature remains below the onset of nucleate boiling. Furthermore, the utilization of Wilks' theorem can reduce computational cost for uncertainty quantification, and the effect of sampling methods is negligible in Monte Carlo simulations with large sample sizes. This approach can enhance the verification and validation (V&V) of regulatory requirements in the licensing process of new reactors.