Summary: | 博士 === 國立臺灣大學 === 機械工程學研究所 === 90 === A probabilistic failure analysis model for the prediction of life and failure probability of a piping system or its components is proposed in the present study. The model incorporates probability and statistical theories into material failure criteria by considering parameters that can affect the propagation of the crack as random and evaluate the failure probability of the pipe. By comparing the results of some numerical examples with those from running the well-known PRAISE Code, it indicates that the proposed model possess a rather good prediction capacity, and in the degree of analysis and calculation time. Based on similar theories, this simpler model can be extended to make a more realistic analysis of a nuclear pipe weld having known initial crack size.
The result of the above probabilistic failure analysis model can also be used in combination with that of the probabilistic risk assessment model. Risk-based preventive maintenance policies for piping systems of either boiling water reactors or pressurized water reactors can therefore be developed. The major concept is to quantify the likely risks existing in the piping systems in order to distinguish their individual importance according to the risk significances of the systems. Based on the degree of importance, the extent and period of plant maintenance can be adjusted rationally. The results of the numerical analysis suggest that the risk-based preventive maintenance policy developed in the present study does have the dual advantages of reducing maintenance cost and still preserving the required plant safety.
|