Study On The Uncertainty In On-line Risk Monitor Of Nuclear Power Plant

On-Line Risk Monitor (OLRM) belonging to the application of Risk-Informed is a technology to evaluate the instantaneous risk of nuclear power plant which is based on the current nuclear power plant configuration, which is used to determine the risk level of the current nuclear power plant operation status and assist operators or administrators making best decisions. OLRM reduces or avoids the high-risk operation, so it improves the defense in depth level and improves security of nuclear power plant. However, owing to the random failure of equipment or system, the distribution of risk cannot be sure as well as the error caused by lacking the knowledge, in other words, the risk calculated from OLRM exists uncertainty. If neglecting the influence of uncertainty, the rationality of the decisions will be questioned. Therefore, it is very important for the application of OLRM to control the source and size of uncertainty.Uncertainty in probabilistic safety assessment of nuclear power plant has been divided into the parameter uncertainty and the model uncertainty, and this forms the theoretical basis of uncertainty. Compared with probabilistic safety assessment,OLRM raises new calculation requirements, so how to achieve the rapid calculation and analysis the new uncertainty caused by the progress of calculation are the main research content of this paper.The calculation of the uncertainty In OLRM needs to meet two requirements: (1)needing to quickly establish the mathematical model of risk for the on-line risk model changing with time; (2) needing to achieve the rapid calculation in order to obtain the uncertainty of the risk.The on-line risk model changes with the time, which represents the change of nuclear power plants configuration. OLRM not only requires the rapid establishment of risk mathematical model, but also need to quickly calculate the risk and risk uncertainty when the configuration changes. This paper develops a Binary Decision Diagram (BDD) algorithm for the rapid establishment of an accurate mathematical model of risk, which is based on modularization techniques. In order to meet the requirements of calculation, it has used three approximate treatments, which are pseudo module, truncation and absorption. In addition, the model uncertainty caused by these approximate treatments is analyzed.The calculation of parameter uncertainty rises with the increasing of the number of parameters and the complexity of the risk model. Therefore, as far as possible to make the results accurate and reduce the calculation,a fast method based on Monte Carlo is proposed.The method used Latin Hypercube Sampling as the uncertainty propagation method, and in order to reduce the number of parameters the method ignores the uncertainty of success events and only takes the 99% contribution of Minimal Cut Sets into account and the method select an appropriate sampling times to make the accuracy and improve the calculation speed.The result shows that the software module development of uncertainty calculation not only can achieve accurate parameters, but also meet the requirements of the calculation speed.Uncertainty study in OLRM defines the main source of uncertainty and achieves the rapid calculation with making the distribution of the risk value clear, which is important for the engineering application of OLRM.