Study On Reliability Of Mechanical Seal In Nuclear Main Pump Based On Zero-Failure Data

The reactor coolant pump in the pressurized water reactor nuclear power plant belongs to the nuclear class I pump,which is called the "heart" of the reactor.Shaft end mechanical seal is a key component of the nuclear main pump,and its reliability has an important impact on the safe and stable operation in nuclear main pumps.In the case of zero-failure data,this thesis effectively combined the grey prediction theory with the optimal confidence limit method to discuss the reliability of the mechanical seal of the nuclear main pump.Based on Bayes theory,the influence of parameter c on the reliability of mechanical seal of nuclear main pump was studied when the prior distribution was Beta distribution.The Bootstrap method was used to solve the reliability estimation interval problem of mechanical seals in nuclear main pumps.First of all,the operational data of the mechanical seal of the Daya Bay nuclear main pump was collected and its reliability distribution was determined,which laid the foundation for the subsequent reliability analysis of the mechanical seal of the nuclear main pump without failure data.Secondly,the censoring time was calculated by using MATLAB software,and the reliability analysis model combining the grey prediction theory and the optimal confidence limit method was established to predict the reliability of the mechanical seal of the nuclear main pump in the case of zero-failure data.Then,a reliability analysis model based on Bayes theory was established.The Monte Carlo method was used to simulate the zero-failure data from the determined reliability distribution.The influence of the value of parameter c on the accuracy of E-Bayes estimation and multi-layer Bayes estimation was discussed when the prior distribution was Beta distribution in the case of zero-failure data.Finally,the reliability interval estimation of mechanical seal of nuclear main pump based on Bootstrap method was studied.The research results can provide theoretical support for the safe operation of the mechanical seal of the nuclear main pump.The main research work and conclusions of this thesis were as follows:(1)The reliability distribution type of the mechanical seal of Daya Bay nuclear main pump was determined.By fitting the data with common different reliability distributions,the goodnessof-fit was calculated by the least square method to determine the type of optimal reliability degree distribution,Weibull distribution with shape parameter m=1.8 and scale parameter ?=12632 was finally determined,which provided a reference for subsequent research on the reliability evaluation of mechanical seals of nuclear main pumps based on zero-failure data.(2)The reliability analysis method of the mechanical seal in the main pump based on the optimal confidence limit method combined with the grey prediction theory was discussed.To solve the reliability evaluation problem of the mechanical seal in the nuclear main pump under the condition of zero-failure data,the operation data of mechanical seal in Daya Bay nuclear main pump was analyzed and its reliability distribution was determined firstly.Secondly,the censoring time was calculated using MATLAB software and a reliability analysis model combining grey prediction theory and optimal confidence limit method was established.Finally,the reliability of the mechanical seal in the nuclear main pump under the condition of zero-failure data was predicted.The error between the reliability calculation result and the true value under Weibull distribution was compared when the shape parameter range was unknown and the shape parameter range was known.Results demonstrated that the present model could evaluate the reliability of the mechanical seal in the nuclear main pump effectively.Besides,under the same confidence degree,the relative error between the predicted reliability and the true reliability when the shape parameter range was known reduced about 10% compared with that when the shape parameter range was unknown.The research results provide guidance for the reliability analysis of the nuclear main pump mechanical seals under the condition of zero-failure data.(3)The reliability analysis method of nuclear main pump mechanical seal based on distribution curve method was discussed.Combined with the idea of distribution curve method,the reliability evaluation model of the mechanical seal in nuclear main pump combined with Bayes theory was established.The Monte Carlo method was used to simulate the zero-failure data from the determined reliability distribution.This thesis discussed the influence of the value of parameter c on the accuracy of E-Bayes estimation and multi-layer Bayes estimation when the prior distribution was Beta distribution in the case of zero-failure data.Four estimation methods of failure probability were compared,and the advantages and disadvantages of different methods were revealed by comparing the calculated results of different methods with the true values.The research showed that the estimation accuracy of the classical method was better than that of the traditional Bayes estimation method.For multi-layer Bayes and E-Bayes estimation,when c<8,multi-layer Bayes estimation was preferred;when c>8,E-Bayes estimation was preferred;when c=8,the average relative error of the two methods reached a lower level and the multi-layer Bayes estimation was lower.Compared with confidence limit method,Bayesian methods have higher reliability prediction accuracy.(4)Due to the fact that the zero-failure data of nuclear main pump mechanical seals generally have the characteristics of small samples,and in recent years,the Bootstrap method has been widely used in small sample data processing and interval estimation problems.This thesis discussed the general analysis steps of the Bootstrap method for estimating the reliability interval of nuclear main pump mechanical seals in zero-failure situations.The research shows that it is feasible to use the Bootstrap method to estimate the reliability of the mechanical seal in nuclear main pump.From the interval length of calculation results,the classical method was estimated to be less stable than Bayesian method,and the stability of the traditional Bayesian method was better than that of other methods.From the estimation results of shape parameters and scale parameters,the fluctuation was large,mainly because the number of self-help samples obtained was not enough.The Bootstrap method is relatively stable in the interval estimation results obtained from processing the zero-failure data,and has certain feasibility.