Algorithm Study And Implementation Of Accident Sequence Quantification In Nuclear Power Plant Probabilistic Safety Assessment

Energy issue is one of the most severe challenges in the world today. It is an important way to solve the energy issue to develop new energy sources, which is represented by nuclear energy. The development of nuclear power cannot materialize without safety, while the safety assessment method is an important means to study safety issues in nuclear power plants (NPPs). Probabilistic Safety Assessment (PSA), which is an important measure in NPPs’safety assessment, is receiving increasingly significant attention. However, there exists calculation performance bottleneck in accident sequence quantification in PSA.The work of this thesis is based on the reliability and PSA software platform RiskA, former developed by the Institute of Nuclear Energy Safety Technology of Chinese Academy of Sciences-FDS Team, and mainly focused on the key points during pre-processing and post-processing in accident sequence quantification. The most important contributions are listed as follows:(1) a complex switching event set processing algorithm based on fast switching set structure detection and structure reuse is proposed, which improves the analysis efficiency of PSA models with complex switching event sets, and significantly enhances the efficiency especially for the joint analysis of multiple sequences of large models. By using this algorithm the average accident sequences quantification time is dramatically reduced; (2) a logical success branch fast treatment algorithm based on cut sets deducing technology is proposed, which has the advantages of wider application range and higher running speed compared with the traditional logical success branch treatment algorithm based on cut sets matching technology, and significantly improves the processing speed of logical success branches in the case of a large number of cut sets,. By using this algorithm the average accident sequences quantification velocity increases significantly; (3) a fault tree quantification algorithm combining zero-suppressed binary decision diagrams (ZBDD) method and binary decision diagrams (BDD) method is proposed, which overcomes excessive consumption of calculation resources in the ZBDD method and lack of calculation accuracy in the BDD method. The algorithm, by combing the advantages of the both, reduces the memory consumption compared with the native BDD method, and increases the top event failure probability accuracy compared with the classical ZBDD method.In addition, the corresponding accident sequence quantification optimization modules are developed on the basis of above research, which are further verified by a number of PSA models of authentic NPPs. The outputs show that the quantification results are accurate and reliable while the analysis velocity is faster than the internationally known software RiskSpectrum.*...