A GO-FLOW and Dynamic Bayesian Network Combination Approach for Reliability Evaluation with Uncertainty: A Case Study on a Nuclear Power Plant

Uncertainty analyses have been considered critical analysis methods for identifying the risks in reliability evaluations. However, with multi-phase, multi-state, and repairable features, this method cannot effectively and precisely display the reliability evaluation results with uncertainty for dynamic and complex systems. In this paper, uncertainty analysis has been conducted in the evaluation of safety-related risk analysis for a nuclear power plant (NPP). A GO-FLOW and dynamic Bayesian network (DBN) combination approach for the reliability evaluation with uncertainty is proposed in this paper. Based on the unified rules, the various operators can be mapped into the DBN even with the multi-phase, multi-state, and repairable characteristics. As the framework of the DBN, utilizing sensitivity analysis, this approach can provide information on those inputs that are contributing the most to the uncertainty. Next, the DBN algorithm and the Monte Carlo simulation are used to quantify the uncertainty in terms of appropriate estimates for the analysis results. Finally, the auxiliary power supply system of the pressurized water reactor in the NPP is analyzed as an example to illustrate the approach. The results of this paper show that uncertainty analysis makes the reliability evaluation more accurate compared with the results without the uncertainty analysis. Moreover, the GO-FLOW methodology can be applied easily for uncertainty analysis with its modified functions and algorithms.