Reliability evaluation of auxiliary feedwater system by mapping GO-FLOW models into Bayesian networks

Area of Science:

• Reliability Engineering
• Probabilistic Systems Analysis
• Computer Science

Background:

• Bayesian networks (BNs) are established for modeling uncertainty in probabilistic systems.
• The GO-FLOW method is a success-oriented technique for system reliability and risk evaluation.
• Limitations exist in current GO-FLOW method applications for BN model development.

Purpose of the Study:

• To present a novel approach for constructing Bayesian networks from GO-FLOW models.
• To develop equivalent BNs for the fourteen basic operators within the GO-FLOW methodology.
• To enable the mapping of existing GO-FLOW models into equivalent BN representations.

Main Methods:

• Development of equivalent Bayesian networks for each of the fourteen basic GO-FLOW operators.
• Establishing a mapping methodology from GO-FLOW models to equivalent BNs based on operator equivalences.
• Application and validation using a case study of an auxiliary feedwater system in a pressurized water reactor.

Main Results:

• Successful development of equivalent BNs for GO-FLOW operators.
• Demonstration of the equivalence between GO-FLOW models and the developed BNs at discrete time points.
• Validation of the proposed mapping method through a practical engineering case study.

Conclusions:

• The proposed method provides a novel approach to construct Bayesian networks from GO-FLOW models.
• The developed Bayesian networks are equivalent to GO-FLOW models at a static time point.
• The successful mapping of GO-FLOW charts to equivalent BNs is confirmed, offering a new tool for reliability and risk analysis.