Research on automatic fault tree construction for nuclear power plants

The PSA model is a crucial industry tool for understanding and evaluating the safety of nuclear power plants. Given the complexity of nuclear systems, the development and maintenance of such models are highly time-consuming and labor-intensive tasks. This paper proposed a fault tree generation method tailored to the nuclear power plants. Utilizing P&IDs and FMEA as inputs, the method applied a tracing algorithm to automatically construct fault trees. Additionally, for special cases in nuclear power plants, such as reverse flow tracing, branch isolation failure handling, and common cause failure group generation, specific solutions were provided. In the final section, we introduced an automatic fault tree modeling tool, AutoFAT, which is characterized by its user-friendly interface and high level of automation. We imported the automatically constructed fault trees into the reliability analysis software RiskSpectrum PSA and compared them to manual modeling results, which showed a time savings of over 70% for modeling individual top events and a translation accuracy of over 90%. Specifically, accuracy was evaluated by comparing basic events, CCF events, calculated top event unavailability, and the minimal cut sets. It is believed that the application of AutoFAT will facilitate PSA modeling, model maintenance and model review.