Incorporation of Passive System Functional Reliability in a Probabilistic Safety Assessment

Advanced nuclear reactor designs incorporate safety passive systems in addition to active ones. Assessment of the capability of such passive systems in performing satisfactorily their functions, also called "functional reliability" is an important issue for the safety analysis. Operating of passive system functions is based on thermohydraulic (T-H) principles, which are not generally considered to be subject to any kind of failures in conventional probabilistic safety assessment (PSA). But due to the environment and to the physical phenomena that may deviate from expectation, passive systems may fail to meet their required functions. Quantification process of the unreliability figures of passive systems have to consider uncertainties of numerous parameters which can affect the accomplishment of system functions during a transient. RMPS (Reliability Method for Passive System) is one of the methodologies that address the reliability assessment of a passive system and its incorporation into a PSA model. Thus, PSA insights will provide risk contributions of mechanical, electrical and I&C system failures, human errors, and functional reliability of passive systems. RMPS methodology uses a statistical approach to assess the reliability of a passive system taking into account the variation uncertainties of the key physical and code parameters. Based on the probability distribution of each main parameter, a Monte Carlo simulation drawing gives a set of randomized value for each parameter. The paper presents a use case performed by EDF on a SAfety COndenser (SACO), a passive emergency feedwater system, in a loss of main feedwater scenario on a large PWR. To incorporate the passive system functional reliability, which is highly dependent on physical phenomena, in the PSA event tree, the analyst considers all the possible T-H states that can be reached by the reactor depending on the passive system performances. Then, for each identified T-H state, all the possible protection and mitigation systems, as well as human actions that can be credited are identified. RMPS results correspond to a probability distribution for each T-H state, and for all relevant functional sequences of the event tree. In this way a comprehensive event tree development is possible by integrating these probabilities as a multiple-branch event. The results of the use case are discussed, and several ways of optimization are presented in an effort to reduce the overall number of sequences, to increase the model readability and to possibly reduce the calculation time.