Research On The Method Of Nuclear Power Plant Accident Risk Assessment And Uncertainty Quantification During Low Power And Shutdown Conditions

Safety is the lifeline of nuclear industry,some previous research and plant operation experiences in recent years have proven that the risk during low power and shutdown conditions of nuclear power plants has the similar magnitude to that of power operation.Thus,performing risk assessment on the low power and shutdown conditions of nuclear power plants contributes to improve the safety and the competitiveness of nuclear power plants.However,there is little research on the method of risk assessment for low power and shutdown conditions,and the alternative way is to modify the traditional probabilistic risk assessment method with some simplifications and assumptions.Obviously,those simplifications and assumptions would introduce extra uncertainties to the risk assessment result and reduce its value,thus this method has not been approved widely.Therefore,this research aims at developing a new risk assessment method that is appropriate for low power and shutdown conditions,and a series of work have been performed.The main contents and the innovativeness are described as follows:(1)To deal with the dynamic characteristics of the plant states during low power and shutdown conditions,a plant configuration-based plant operation states(POS)definition method is proposed,which deal with the variation of model parameters as the uncertainties.Subsequently,the plant model during each POS is modeled by the best estimated code.With these plant models,accident sequence and mitigation strategies are analyzed,and some special risk for low power and shutdown conditions,such as the inadequate of severe accident management guideline entry condition and the hydrogen risk in some apartments of containment,are also identified.(2)A risk-informed risk assessment method for low power and shutdown conditions is introduced,in which the model parameter uncertainties are considered and quantified in the risk assessment model by introducing the concept of risk informed safety margin.An example is also given by applying this method to a reflux cooling related sequence following the loss of residual heat removal accident.(3)Surrogate model-based uncertainty quantification method is introduced to deal with the low efficiency of Monte-Carlo Sampling-based uncertainty quantification in the risk-informed risk assessment method,and to improve the efficiency of uncertainty quantification in the risk assessment.During the modeling processes,an adaptive sampling-based Ada-PC-Kriging method is also developed to inprove the quality of samples and the model accuracy.In addition,the surrogate mode is also used to perform sensitivity analysis and the limit state evaluation.(4)A probability box(p-box)and surrogate model-based uncertainty quantification is proposed to consider the hybrid aleatory uncertainty and epistemic uncertainty by a probability box,and the influence of different type of uncertainty on the system output can be also characterized by a double-level surrogate model-based uncertainty propagation method.In addition,the propragation procedures of uncertainties through the system model can be also optimized by introducing the empirical knowledge.Through the probability box(p-box)and surrogate model-based uncertainty quantification framework,the influence of epistemic uncertainty on the model output can be characterized by the area between the lower and upper bound of p-box,and the lower and upper bound of p-box can be modeled by two boundary function which is only related to the aleatory uncertainties.Thus,the probability box(p-box)and surrogate model-based uncertainty quantification method can reduce the overall model dimension and have higher efficiency when applied to cases with more uncertainties.With the research contents above,a series of full-scope risk assessment and uncertainty quantification method for low power and shutdown conditions of nuclear power plants,which has higher efficiency,stronger universality,and less uncertainties with respect to the traditional probabilistic risk assessment method.Thus,the method developed in this research can be utilized in various type of nuclear power plants to quantify the risk during low power and shutdown conditions accurately.