Dynamic accident sequence analysis using dynamic flowgraph method and Markov/cell-to-cell mapping technique

Abstract

In the recent years, numerous concerns have been raised regarding the capabilities and adequacy of classical probabilistic safety assessment (PSA) techniques (fault/event tree) to account for dynamic system interactions and time-dependent accident sequence evolution. Subsequently there is an interest within the PSA community to complement the classical techniques with dynamic methodologies; driven in addition by a goal to eventually develop a framework for integrated safety assessment. The first phase of this research investigates and addresses the limitations of classical techniques and performs a methodological comparison between classical and dynamic PSA techniques. The dynamic flowgraph method (DFM) and Markov model coupled with cell-to-cell mapping technique (Markov-CCMT) were the two dynamic methodologies selected for this research. These methods are ranked as the top methodologies with favorable and minimal uncertainties as determined by the United States Nuclear Regulatory Commission. The capabilities and limitations of the techniques are demonstrated by applying it to a benchmark liquid level control system exhibiting dynamic characteristics and interactions. Reliability analysis of the system using ET/FT are performed using the CAFTA code. DFM model of the benchmark system is developed using the DYMONDA code and coupling of Markov-CCMT model is performed using Fortran95 and MATLAB code. Classical techniques were found to overestimate the predicted top event frequencies by more than one order of magnitude depending on whether or not dynamic interactions among the units through the state variable is accounted for. The study shows that DFM focuses more on sequential probabilistic system evolution, whereas Markov-CCMT emphasizes the exact timing of a failure event. The second phase involves the development of a novel approach for integrated reliability assessment of passive safety systems in small modular reactors. A stochastic model of a passive Isolation Condenser System (ICS) was developed, and its state transition probabilities are computed using finite element method. The analysis predicts high system reliability, with the ICS most likely to fail by pressure boundary breach followed by condensate return and venting unit failure.