Petri nets based reliability modeling of reactor protection system considering periodic surveillance test

Reactor protection system (RPS) is one of the most important safety systems in nuclear power plant (NPP). A RPS usually consists of 4 redundant channels, in which a triggering signal is generated to safely shut down the reactor if necessary. The reliability analysis and assessment of RPS is a very important aspect which has been received widespread concern. Fault tree analysis (FTA) and Markov chain are two conventional reliability analysis methods for RPS. However, FTA is a static methodology which can not model the periodic surveillance test (PST) and maintenance. And Markov chain is only applicable for the exponential distribution process, whereas the failure and maintenance process for RPS do not always follow exponential distribution. Moreover, the RPS will degrade the voting logic from 2-out-of-4 to 2-out-of-3 during the PST. In fact, there are few models consider this degradation process. In this paper, a dynamic reliability model for the RPS is proposed based on Petri nets. The dynamic transition processes of all states are described by Petri nets, in which the PST and maintenance are considered. Different from other models assuming the PST process following an exponential distribution, a deterministic time duration of the PST is adopted in this model. During the PST, the degradation process that voting logic transit from 2-out-of-4 to 2-out-of-3 is determined. The correctness and applicability of the model are analyzed by comparing the analytical solution and numerical solution. A comparison among the other two models and the proposed model is simulated. The result shows that degradation process voting logic from 2-out-of-4 to 2-out-of-3 will significantly increase the instantaneous unavailability during the PST. Therefore, it is indeed necessary to accurately model maintenance process and the degradation process result from that for RPS. (author)