Resilience assessment of safety-critical systems with credal networks

Nuclear accidents like Fukushima Daiichi or Three Mile Island have changed the safety paradigm to address not only the system failure but also the factors for safety. New nuclear power plant designs should consider safety parameters that maintain the operating conditions while coping with uncertain and possibly disruptive events. In other words, safety systems should increase the resilience of the nuclear power plants. In order to do so, resilience assessments must be carried out to study the system performance recovery once a disruptive event has occurred. The general process of resilience engineering comprises four states changing over time: reliability, unreliability, recovery and recovered steady-state. Probabilistic models like fault tree analysis, have been widely applied in the nuclear industry (among others) due to their flexibility to model complex engineering systems and uncertainty quantification of failure probabilities. However, such technique moderate the modeling scope when representing the interdependencies of the components in the system and variations in time over a disruption event. Moreover, additional complications in the resilience assessment process arise when considering the epistemic uncertainty due to the lack of knowledge about the events and the operating conditions. In this work, credal networks have been proposed for modeling complex interconnected systems as well as taking into account the epistemic uncertainty attached to the lack of data. The resilience assessment of the Main Heat Transport System together with key safety systems of an Advanced Thermal Reactor is carried out to evaluate the system recovery after a mishap adopting the credal network approach. The case study considers a heat removal system of a typical thermal reactor under high-pressure transients. This disruption event would activate the passive safety system to fulfill the reactor shut down process. The application of the proposed approach to producing a resilience analysis is described and results presented to demonstrate the applicability of the method.